186 KiB
Potatoes and Portal Guns
CLOSED: [2011-04-26 Tue]
Meh.php
CLOSED: [2011-04-27 Wed 00:00]
Transmission, RSS, and XBMC
CLOSED: [2011-04-27 Wed 00:01]
Learning Functional Programming, Part One
CLOSED: [2012-04-09 Mon]
Erlang: The Movie
CLOSED: [2013-11-27 Wed]
Getting Organized with Org Mode
CLOSED: [2014-11-25 Tue]
I've been using Emacs Org mode for nearly a year now. For a while I mostly just used it to take and organize notes, but over time I've discovered it's an incredibly useful tool for managing projects and tasks, writing and publishing documents, keeping track of time and todo lists, and maintaining a journal.
Project Management
Most of what I've been using Org mode for has been breaking down large projects at work into tasks and subtasks. It's really easy to enter projects in as a hierarchy of tasks and task groupings. Using Column View, I was able to dive right into scoping them individually and reporting total estimates for each major segment of work.
Because Org Mode makes building and modifying an outline structure like this so quick and easy, I usually build and modify the project org document while planning it out with my team. Once done, I then manually load that information into our issue tracker and get underway. Occasionally I'll also update tags and progress status in the org document as well as the project progresses, so I can use the same document to plan subsequent development iterations.
Organizing Notes and Code Exercises
More recently, I've been looking into various ways to get more things organized with Org mode. I've been stepping through Structure and Interpretation of Computer Programs with some other folks from work, and discovered that Org mode was an ideal fit for keeping my notes and exercise work together. The latter is neatly managed by Babel, which let me embed and edit source examples and my excercise solutions right in the org document itself, and even export them to one or more scheme files to load into my interpreter.
Exporting and Publishing Documents
Publishing my notes with org is also a breeze. I've published project plans and proposals to PDF to share with colleagues, and exported my SICP notes to html and dropped them into a site built with Jekyll. Embedding graphs and diagrams into exported documents using Graphviz, Mscgen, and PlantUML has also really helped with putting together some great project plans and documentation. A lot of great examples using those tools (and more!) can be found here.
Emacs Configuration
While learning all the cool things I could do with Org mode and Babel, it was only natural I'd end up using it to reorganize my Emacs configuration. Up until that point, I'd been managing my configuration in a single init.el file, plus a directory full of mode or purpose-specific elisp files that I'd loop through and load. Inspired primarily by the blog post, "Making Emacs Work For Me", and later by others such as Sacha Chua's Emacs configuration, I got all my configs neatly organized into a single org file that gets loaded on startup. I've found it makes it far easier to keep track of what I've got configured, and gives me a reason to document and organize things neatly now that it's living a double life as a published document on GitHub. I've still got a directory lying around with autoloaded scripts, but now it's simply reserved for tinkering and sensitive configuration.
Tracking Habits
Another great feature of Org mode that I've been taking advantage of a lot more lately is the Agenda. By defining some org files as being agenda files, Org mode can examine these files for TODO entries, scheduled tasks, deadlines and more to build out useful agenda views to get a quick handle on what needs to be done and when. While at first I started by simply syncing down my google calendars as org-files (using ical2org.awk), I've started managing TODO lists in a dedicated org file. By adding tasks to this file, scheduling them, and setting deadlines, I've been doing a much better job of keeping track of things I need to get done and (even more importantly) when I need to get them done.
This works not only for one-shot tasks, but also habits and other repetitive tasks. It's possible to schedule a task that should be done every day, every few days, or maybe every first sunday of a month. For example, I've set up repeating tasks to write a blog post at least once a month, practice guitar every two to three days, and to do the dishes every one or two days. The agenda view can even show a small, colorized graph next to each repeating task that paints a picture of how well (or not!) I've been getting those tasks done on time.
Keeping a Journal and Tracking Work
The last thing I've been using (which I'm still getting a handle on) is using Capture to take and store notes, keep a journal, and even track time on tasks at work.
(setq org-capture-templates
'(("j" "Journal Entry" plain
(file+datetree "~/org/journal.org")
"%U\n\n%?" :empty-lines-before 1)
("w" "Log Work Task" entry
(file+datetree "~/org/worklog.org")
"* TODO %^{Description} %^g\n%?\n\nAdded: %U"
:clock-in t
:clock-keep t)))
(global-set-key (kbd "C-c c") 'org-capture)
(setq org-clock-persist 'history)
(org-clock-persistence-insinuate)For my journal, I've configured a capture template that I can use to write down a new entry that will be stored with a time stamp appended into its own org file, organized under headlines by year, month and date.
For work tasks, I have another capture template configured that will log and tag a task into another org file, also organized by date, which will automatically start tracking time for that task. Once done, I can simply clock out and check the time I've spent, and can easily find it later to clock in again, add notes, or update its status. This helps me keep track of what I've gotten done during the day, keep notes on what I was doing at any point in time, and get a better idea of how long it takes me to do different types of tasks.
Conclusion
There's a lot that can be done with Org mode, and I've only just scratched the surface. The simple outline format provided by Org mode lends itself to doing all sorts of things, be it organizing notes, keeping a private or work journal, or writing a book or technical document. I've even written this blog post in Org mode! There's tons of functionality that can be built on top of it, yet the underlying format itself remains simple and easy to work with. I've never been great at keeping myself organized, but Org mode is such a delight to use that I can't help trying anyway. If it can work for me, maybe it can work for you, too!
There's tons of resources for finding new ways for using Org mode, and I'm still discovering cool things I can track and integrate with it. I definitely recommend reading through Sacha Chua's Blog, as well as posts from John Wiegley. I'm always looking for more stuff to try out. Feel free to drop me a line if you find or are using something you think is cool or useful!
Adventuring Through SICP
CLOSED: [2015-01-01 Thu]
Back in May, a coworker and I got the idea to start up a little seminar after work every couple of weeks with the plan to set aside some time to learn and discuss new ideas together, along with anyone else who cared to join us.
Learning Together
Over the past several months, we've read our way through the first three chapters of the book, watched the related video lectures, and did (most of) the exercises.
Aside from being a great excuse to unwind with friends after work (which it is!), it's proved to be a great way to get through the material. Doing a section of a chapter every couple of weeks is an easy goal to meet, and meeting up to discuss it becomes something to look forward to. We all get to enjoy a sense of accomplishment in learning stuff that can be daunting or difficult to set aside time for alone.
The best part, by far, is getting different perspectives on the material. Most of my learning tends to be solitary, so it's refreshing to do it with a group. By reviewing the different concepts together, we're able to gain insights and clarity we'd never manage on our own. Even the simplest topics can spur interesting conversations.
SICP
Our first adventure together so far has been the venerable Structure and Interpretation of Computer Programs. This book had been on my todo list for a long time, but never quite bubbled to the top. I'm glad to have the opportunity to go through it in this format, since there's plenty of time to let really get into the excercises and let the lessons sink in.
SICP was originally an introductory textbook for MIT computer programming courses. What sets it apart from most, though, is that it doesn't focus so much on learning a particular programming language (while the book does use and cover MIT Scheme) as it does on identifying and abstracting out patterns common to most programming problems. Because of that, the book is every bit as useful and illuminating as ever, especially now that functional paradigms are re-entering the spotlight and means of abstracting and composing systems are as important as ever.
What's next?
We've still got plenty of SICP left to get through. We've only just gotten through Chapter 4, section 1, which has us building a scheme interpreter in scheme, so there's plenty of fun left to be had there.
We're also staring to do some smaller, lunchtime review meetings following the evening discussions to catch up the folks that can't make it. I may also try sneaking in some smaller material, like interesting blog posts, to keep things lively.
If anyone's interested, I have the exercise work along with some notes taken during the meetings hosted online. I apologize for the lack of notes early on, I've been trying to get better at capturing memorable excerpts and conversation topics recently. I may have to put some more posts together later on summarizing what we discussed for each chapter; if and when I do, they'll be posted on the seminar website.
Coders at Work
CLOSED: [2015-01-28 Wed]
A few days before leaving work for a week and a half of flying and cruising to escape frigid Pennsylvania, I came across a Joe Armstrong quote during my regularly scheduled slacking off on twitter and Hacker News. I'd come across a couple times before, only this time I noticed it had a source link. This led me to discovering (and shortly thereafter, buying) Peter Seibel's "Coders at Work – Reflections on the Craft of Programming". I loaded it onto my nook, and off I went.
The book is essentially a collection of interviews with a series of highly accomplished software developers. Each of them has their own fascinating insights into the craft and its rich history.
While making my way through the book, I highlighted some excerpts that, for one reason or another, resonated with me. I've organized and elaborated on them below.
Incremental Changes
CLOSED: [2015-01-20 Tue 20:59] <<fitzpatrick-increments>>
I've seen young programmers say, "Oh, shit, it doesn't work," and then rewrite it all. Stop. Try to figure out what's going on. Learn how to write things incrementally so that at each stage you could verify it.
– Brad Fitzpatrick
I can remember doing this to myself when I was still relatively new to coding (and even worse, before I discovered source control!). Some subroutine or other would be misbehaving, and rather than picking it apart and figuring out what it was I'd done wrong, I'd just blow it away and attempt to write it fresh. While I might be successful, that likely depended on the issue being some sort of typo or missed logic; if it was broken because I misunderstood something or had a bad plan to begin with, rewriting it would only result in more broken code, sometimes in more or different ways than before. I don't think I've ever rewritten someone else's code without first at least getting a firm understanding of it and what it was trying to accomplish, but even then, breaking down changes piece by piece makes it all the easier to maintain sanity.
I do still sometimes catch myself doing too much at once when building a new feature or fixing a bug. I may have to fix a separate bug that's in my way, or I may have to make several different changes in various parts of the code. If I'm not careful, things can get out of hand pretty quickly, and before I know it I have a blob of changes strewn across the codebase in my working directory without a clear picture of what's what. If something goes wrong, it can be pretty tough to sort out which change broke things (or fixed them). Committing changes often helps tremendously to avoid this sort of situation, and when I catch myself going off the rails I try to find a stopping point and split changes up into commits as soon as possible to regain control. Related changes and fixes can always be squashed together afterwards to keep things tidy.
Specifications & Documentation
CLOSED: [2015-01-20 Tue 20:59] <<bloch-customers>>
Many customers won't tell you a problem; they'll tell you a solution. A customer might say, for instance, "I need you to add support for the following 17 attributes to this system. Then you have to ask, 'Why? What are you going to do with the system? How do you expect it to evolve?'" And so on. You go back and forth until you figure out what all the customer really needs the software to do. These are the use cases.
– Joshua Bloch
Whether your customer is your customer, or your CEO, the point stands: customers are really bad at expressing what they want. It's hard to blame them, though; analyzing what you really want and distilling it into a clear specification is tough work. If your customer is your boss, it can be intimidating to push back with questions like "Why?", but if you can get those questions answered you'll end up with a better product, a better understanding of the product, and a happy customer. The agile process of doing quick iterations to get tangible results in front of them is a great way of getting the feedback and answers you need.
<<armstrong-documentation>>
The code shows me what it does. It doesn't show me what it's supposed to do. I think the code is the answer to a problem. If you don't have the spec or you don't have any documentation, you have to guess what the problem is from the answer. You might guess wrong.
– Joe Armstrong
Once you've got the definition of what you've got to build and how it's got to work, it's extremely important that you get it documented. Too often, I'm faced with code that's doing something in some way that somebody, either a customer or a developer reading it, takes issue with, and there's no documentation anywhere on why it's doing what it's doing. What happens next is anybody's guess. Code that's clear and conveys its intent is a good start towards avoiding this sort of situation. Comments explaining intent help too, though making sure they're kept up to date with the code can be challenging. At the very least, I try to promote useful commit messages explaining what the purpose of a change is, and reference a ticket in our issue tracker which (hopefully) has a clear accounting of the feature or bugfix that prompted it.
Pair Programming
CLOSED: [2015-01-20 Tue 21:03] <<armstrong-pairing>>
… if you don't know what you're doing then I think it can be very helpful with someone who also doesn't know what they're doing. If you have one programmer who's better than the other one, then there's probably benefit for the weaker programmer or the less-experienced programmer to observe the other one. They're going to learn something from that. But if the gap's too great then they won't learn, they'll just sit there feeling stupid.
– Joe Armstrong
Pairing isn't something I do much. At least, it's pretty rare that I have someone sitting next to me as I code. I do involve peers while I'm figuring out what I want to build as often as I can. The tougher the problem, the more important it is, I think, to get as much feedback and brainstorming in as possible. This way, everybody gets to tackle the problem and learn together, and anyone's input, however small it might seem, can be the key to the "a-ha" moment to figuring out a solution.
Peer Review
CLOSED: [2015-01-25 Sun 22:44] <<crockford-reading>>
I think an hour of code reading is worth two weeks of QA. It's just a really effective way of removing errors. If you have someone who is strong reading, then the novices around them are going to learn a lot that they wouldn't be learning otherwise, and if you have a novice reading, he's going to get a lot of really good advice.
– Douglas Crockford
Just as important as designing the software as a team, I think, is reviewing it as a team. In doing so, each member of the team has an opportunity to understand how the system has been implemented, and to offer their suggestions and constructive criticisms. This helps the team grow together, and results in a higher quality of code overall. This benefits QA as well as the developers themselves for the next time they find themselves in that particular bit of the system.
Object-Oriented Programming
CLOSED: [2015-01-20 Tue 20:59] <<armstrong-oop>>
I think the lack of reusability comes in object-oriented languages, not in functional languages. Because the problem with object-oriented languages is they've got all this implicit environment that they carry around with them. You wanted a banana but what you got was a gorilla holding the banana and the entire jungle.
– Joe Armstrong
A lot has been written on why OOP isn't the great thing it claims to be, or was ever intended to be. Having grappled with it myself for years, attempting to find ways to keep my code clean, concise and extensible, I've more or less come to the same conclusion as Armstrong in that coupling data structures with behaviour makes for a terrible mess. Dividing the two led to a sort of moment of clarity; there was no more confusion about what methods belong on what object. There was simply the data, and the methods that act on it. I am still struggling a bit, though, on how to bring this mindset to the PHP I maintain at work. The language seems particularly ill-suited to managing complex data structures (or even simple ones – vectors and hashes are bizarrely intertwined).
Writing
CLOSED: [2015-01-28 Wed 22:42] <<bloch-writing>>
You should read [Elements of Style] for two reasons: The first is that a large part of every software engineer's job is writing prose. If you can't write precise, coherent, readable specs, nobody is going to be able to use your stuff. So anything that improves your prose style is good. The second reason is that most of the ideas in that book are also applicable to programs.
– Joshua Bloch
<<crockford-writing>>
My advice to everybody is pretty much the same, to read and write.
…
Are you a good Java programmer, a good C programmer, or whatever? I don't care. I just want to know that you know how to put an algorithm together, you understand data structures, and you know how to document it.
– Douglas Crockford
<<knuth-writing>>
This is what literate programming is so great for –
I can talk to myself. I can read my program a year later and know exactly what I was thinking.
– Donald Knuth
The more I've program professionally, the clearer it is that writing (and communication in general) is a very important skill to develop. Whether it be writing documentation, putting together a project plan, or whiteboarding and discussing something, clear and concise communication skills are a must. Clarity in writing translates into clarity in coding as well, in my opinion. Code that is short, to the point, clear in its intention, making good use of structure and wording (in the form of function and variable names) is far easier to read and reason about than code that is disorganized and obtuse.
Knuth
CLOSED: [2015-01-28 Wed 22:42] <<crockford-knuth>>
I tried to make familiarity with Knuth a hiring criteria, and I was disappointed that I couldn't find enough people that had read him. In my view, anybody who calls himself a professional programmer should have read Knuth's books or at least should have copies of his books.
– Douglas Crockford
<<steele-knuth>>
… Knuth is really good at telling a story about code. When you read your way through The Art of Computer Programming and you read your way through an algorithm, he's explained it to you and showed you some applications and given you some exercises to work, and you feel like you've been led on a worthwhile journey.
– Guy Steele
<<norvig-knuth>>
At one point I had [The Art of Computer Programming] as my monitor stand because it was one of the biggest set of books I had, and it was just the right height. That was nice because it was always there, and I guess then I was more prone to use it as a reference because it was right in front of me.
– Peter Norvig
I haven't read any of Knuth's books yet, which is something I'll have to rectify soon. I don't think I have the mathematical background necessary to get through some of his stuff, but I expect it will be rewarding nonetheless. I'm also intrigued by his concept of literate programming, and I'm curious to learn more about TeX. I imagine I'll be skimming through TeX: The Program pretty soon now that I've finished Coders at Work :)
Birthday Puzzle
CLOSED: [2015-04-18 Sat]
This logic puzzle has been floating around the internet lately. When I caught wind of it, I thought it would be a great exercise to tackle using Prolog. I'm not especially good with the language yet, so it added to the challenge a bit, but it was a pretty worthwhile undertaking. When I got stumped, I discovered that mapping out the birthdays into a grid helped me visualize the problem and ultimately solve it, so I've included that with my prolog code so you can see how I arrived at the answer.
The Puzzle
Albert and Bernard have just met Cheryl. “When is your birthday?” Albert asked Cheryl. Cheryl thought for a moment and said, “I won’t tell you, but I’ll give you some clues”. She wrote down a list of ten dates:
- May 15, May 16, May 19
- June 17, June 18
- July 14, July 16
- August 14, August 15, August 17
“One of these is my birthday,” she said.
Cheryl whispered in Albert’s ear the month, and only the month, of her birthday. To Bernard, she whispered the day, and only the day. “Can you figure it out now?” she asked Albert.
Albert: “I don’t know when your birthday is, but I know Bernard doesn’t know, either.”
Bernard: “I didn’t know originally, but now I do.”
Albert: “Well, now I know, too!”
When is Cheryl’s birthday?
The Solution
The Dates
To start off, i entered each of the possible birthdays as facts:
possible_birthday(may, 15).
possible_birthday(may, 16).
possible_birthday(may, 19).
possible_birthday(june, 17).
possible_birthday(june, 18).
possible_birthday(july, 14).
possible_birthday(july, 16).
possible_birthday(august, 14).
possible_birthday(august, 15).
possible_birthday(august, 17).And here they are, mapped out in a grid:
| May | June | July | August | |
|---|---|---|---|---|
| 14 | X | X | ||
| 15 | X | X | ||
| 16 | X | X | ||
| 17 | X | X | ||
| 18 | X | |||
| 19 | X |
Albert's Statement
I don’t know when your birthday is,…
Albert only knows the month, and the month isn't enough to uniquely identify Cheryl's birthday.
month_is_not_unique(M) :-
bagof(D, possible_birthday(M, D), Days),
length(Days, Len),
Len > 1.… but I know Bernard doesn’t know, either.
Albert knows that Bernard doesn't know Cheryl's birthday. Therefore, the day alone isn't enough to know Cheryl's birthday, and we can infer that the month of Cheryl's birthday does not include any of the unique dates.
day_is_not_unique(D) :-
bagof(M, possible_birthday(M, D), Months),
length(Months, Len),
Len > 1.
month_has_no_unique_days(M) :-
forall(possible_birthday(M,D),
day_is_not_unique(D)).Based on what Albert knows at this point, let's see how we've reduced the possible dates:
part_one(M,D) :-
possible_birthday(M,D),
month_is_not_unique(M),
month_has_no_unique_days(M),
day_is_not_unique(D).Results = [ (july, 14), (july, 16), (august, 14), (august, 15), (august, 17)].
So the unique days (the 18th and 19th) are out, as are the months that contained them (May and June).
| July | August | |
|---|---|---|
| 14 | X | X |
| 15 | X | |
| 16 | X | |
| 17 | X |
Bernard's Statement
I didn’t know originally, but now I do.
For Bernard to know Cheryl's birthday, the day he knows must be unique within the constraints we have so far.
day_is_unique(Month, Day) :-
findall(M, part_one(M, Day), [Month]).
part_two(Month, Day) :-
possible_birthday(Month, Day),
day_is_unique(Month, Day).Results = [ (july, 16), (august, 15), (august, 17)].
Both July and August contain the 14th, so that row is out.
| July | August | |
|---|---|---|
| 15 | X | |
| 16 | X | |
| 17 | X |
Albert's Second Statement
Well, now I know, too!
Albert's month must be the remaining unique month:
month_is_not_unique(Month, Day) :-
findall(D, part_two(Month, D), [Day]).
part_three(Month, Day) :-
possible_birthday(Month, Day),
month_is_not_unique(Month, Day).Results = [ (july, 16)].
August had two possible days, so it's now clear that the only possible unique answer is July 16th.
| July | |
|---|---|
| 15 | |
| 16 | X |
| 17 |
Cheryl's Birthday
cheryls_birthday(Month, Day) :-
part_three(Month, Day).Month = july, Day = 16.
So, there we have it. Cheryl's birthday is July 16th!
| July | |
|---|---|
| 16 | X |
Keeping Files And Configuration In Sync
CLOSED: [2015-04-20 Mon]
I have a few computers I use on a daily basis, and I like to keep the same emacs and shell configuration on all of them, along with my org files and a handful of scripts. Since I'm sure other people have this problem as well, I'll share what I'm doing so anyone can learn from (or criticise) my solutions.
Git for configuration and projects
I'm a software developer, so keeping things in git just makes sense to me. I keep my org files in a privately hosted git repository, and Emacs and Zsh configurations in a public repo on github. My blog is also hosted and published on github as well; I like having it cloned to all my machines so I can work on drafts wherever I may be.
My .zshrc installs oh-my-zsh if it isn't installed already, and sets up my shell theme, path, and some other environmental things.
My Emacs configuration behaves similarly, making use of John Wiegley's excellent use-package tool to ensure all my packages are installed if they're not already there and configured the way I like them.
All I have to do to get running on a new system is to install git, emacs and zsh, clone my repo, symlink the files, and grab a cup of tea while everything installs.
Bittorrent sync for personal settings & books
For personal configuration that doesn't belong in and/or is too sensitive to be in a public repo, I have a folder of dotfiles and things that I sync between my machines using Bittorrent Sync. The dotfiles are arranged into directories by their purpose:
[correlr@reason:~/dotenv]
% tree -a -L 2
.
├── authinfo
│ └── .authinfo.gpg
├── bin
│ └── .bin
├── emacs
│ ├── .bbdb
│ └── .emacs.local.d
├── mail
│ ├── .gnus.el
│ ├── .signature
├── README.org
├── .sync
│ ├── Archive
│ ├── ID
│ ├── IgnoreList
│ └── StreamsList
├── tex
│ └── texmf
├── xmonad
│ └── .xmonad
└── zsh
└── .zshenv
This folder structure allows my configs to be easily installed using
GNU Stow from my dotenv folder:
stow -vvS *
Running that command will, for each file in each of the directories, create a symlink to it in my home folder if there isn't a file or directory with that name there already.
Bittorrent sync also comes in handy for syncing my growing Calibre ebook collection, which outgrew my Dropbox account a while back.
Drawing Git Graphs with Graphviz and Org-Mode
CLOSED: [2015-07-12 Sun]
Digging through Derek Feichtinger's org-babel examples (which I came across via irreal.org), I found he had some great examples of displaying git-style graphs using graphviz. I thought it'd be a fun exercise to generate my own graphs based on his graphviz source using elisp, and point it at actual git repos.
Getting Started
I started out with the goal of building a simple graph showing a mainline branch and a topic branch forked from it and eventually merged back in.
Using Derek's example as a template, I described 5 commits on a master branch, plus two on a topic branch.
digraph G {
rankdir="LR";
bgcolor="transparent";
node[width=0.15, height=0.15, shape=point, color=white];
edge[weight=2, arrowhead=none, color=white];
node[group=master];
1 -> 2 -> 3 -> 4 -> 5;
node[group=branch];
2 -> 6 -> 7 -> 4;
}The resulting image looks like this:
Designing the Data Structure
The first thing I needed to do was describe my data structure. Leaning on my experiences reading and working through SICP, I got to work building a constructor function, and several accessors.
I decided to represent each node on a graph with an id, a list of parent ids, and a group which will correspond to the branch on the graph the commit belongs to.
(defun git-graph/make-node (id &optional parents group)
(list id parents group))
(defun git-graph/node-id (node)
(nth 0 node))
(defun git-graph/node-parents (node)
(nth 1 node))
(defun git-graph/node-group (node)
(nth 2 node))Converting the structure to Graphviz
Now that I had my data structures sorted out, it was time to step through them and generate the graphviz source that'd give me the nice-looking graphs I was after.
The graph is constructed using the example above as a template. The nodes are defined first, followed by the edges between them.
(defun git-graph/to-graphviz (id nodes)
(string-join
(list
(concat "digraph " id " {")
"bgcolor=\"transparent\";"
"rankdir=\"LR\";"
"node[width=0.15,height=0.15,shape=point,fontsize=8.0,color=white,fontcolor=white];"
"edge[weight=2,arrowhead=none,color=white];"
(string-join
(-map #'git-graph/to-graphviz-node nodes)
"\n")
(string-join
(-uniq (-flatten (-map
(lambda (node) (git-graph/to-graphviz-edges node nodes))
nodes)))
"\n")
"}")
"\n"))For the sake of readability, I'll format the output:
(defun git-graph/to-graphviz-pretty (id nodes)
(with-temp-buffer
(graphviz-dot-mode)
(insert (git-graph/to-graphviz id nodes))
(indent-region (point-min) (point-max))
(buffer-string)))Each node is built, setting its group attribute when applicable.
(defun git-graph/to-graphviz-node (node)
(let ((node-id (git-graph/to-graphviz-node-id
(git-graph/node-id node))))
(concat node-id
(--if-let (git-graph/node-group node)
(concat "[group=\"" it "\"]"))
";")))Graphviz node identifiers are quoted to avoid running into issues with spaces or other special characters.
(defun git-graph/to-graphviz-node-id (id)
(format "\"%s\"" id))For each node, an edge is built connecting the node to each of its parents.
(defun git-graph/to-graphviz-edges (node &optional nodelist)
(let ((node-id (git-graph/node-id node))
(parents (git-graph/node-parents node))
(node-ids (-map #'git-graph/node-id nodelist)))
(-map (lambda (parent)
(unless (and nodelist (not (member parent node-ids)))
(git-graph/to-graphviz-edge node-id parent)))
parents)))
(defun git-graph/to-graphviz-edge (from to)
(concat
(git-graph/to-graphviz-node-id to)
" -> "
(git-graph/to-graphviz-node-id from)
";"))With that done, the simple graph above could be generated with the following code:
(git-graph/to-graphviz-pretty
"example"
(list (git-graph/make-node 1 nil "master")
(git-graph/make-node 2 '(1) "master")
(git-graph/make-node 3 '(2) "master")
(git-graph/make-node 4 '(3 7) "master")
(git-graph/make-node 5 '(4) "master")
(git-graph/make-node 6 '(2) "branch")
(git-graph/make-node 7 '(6) "branch")))Which generates the following graphviz source:
<<git-example()>>The generated image matches the example exactly:
Adding Labels
The next thing my graph needed was a way of labeling nodes. Rather than trying to figure out some way of attaching a separate label to a node, I decided to simply draw a labeled node as a box with text.
digraph G {
rankdir="LR";
bgcolor="transparent";
node[width=0.15, height=0.15, shape=point,fontsize=8.0,color=white,fontcolor=white];
edge[weight=2, arrowhead=none,color=white];
node[group=main];
1 -> 2 -> 3 -> 4 -> 5;
5[shape=box,label=master];
node[group=branch1];
2 -> 6 -> 7 -> 4;
7[shape=box,label=branch];
}
Updating the Data Structure
I updated my data structure to support an optional label applied to a node. I opted to store it in an associative list alongside the group.
(defun git-graph/make-node (id &optional parents options)
(list id parents options))
(defun git-graph/node-id (node)
(nth 0 node))
(defun git-graph/node-parents (node)
(nth 1 node))
(defun git-graph/node-group (node)
(cdr (assoc 'group (nth 2 node))))
(defun git-graph/node-label (node)
(cdr (assoc 'label (nth 2 node))))Updating the Graphviz node generation
The next step was updating the Graphviz generation functions to handle the new data structure, and set the shape and label attributes of labeled nodes.
(defun git-graph/to-graphviz-node (node)
(let ((node-id (git-graph/to-graphviz-node-id (git-graph/node-id node))))
(concat node-id
(git-graph/to-graphviz-node--attributes node)
";")))
(defun git-graph/to-graphviz-node--attributes (node)
(let ((attributes (git-graph/to-graphviz-node--compute-attributes node)))
(and attributes
(concat "["
(mapconcat (lambda (pair)
(format "%s=\"%s\""
(car pair) (cdr pair)))
attributes
", ")
"]"))))
(defun git-graph/to-graphviz-node--compute-attributes (node)
(-filter #'identity
(append (and (git-graph/node-group node)
(list (cons 'group (git-graph/node-group node))))
(and (git-graph/node-label node)
(list (cons 'shape 'box)
(cons 'label (git-graph/node-label node)))))))I could then label the tips of each branch:
(git-graph/to-graphviz-pretty
"labeled"
(list (git-graph/make-node 1 nil '((group . "master")))
(git-graph/make-node 2 '(1) '((group . "master")))
(git-graph/make-node 3 '(2) '((group . "master")))
(git-graph/make-node 4 '(3 7) '((group . "master")))
(git-graph/make-node 5 '(4) '((group . "master")
(label . "master")))
(git-graph/make-node 6 '(2) '((group . "branch")))
(git-graph/make-node 7 '(6) '((group . "branch")
(label . "branch")))))
Automatic Grouping Using Leaf Nodes
Manually assigning groups to each node is tedious, and easy to accidentally get wrong. Also, with the goal to graph git repositories, I was going to have to figure out groupings automatically anyway.
To do this, it made sense to traverse the nodes in topological order.
Repeating the example above,
digraph G {
rankdir="LR";
bgcolor="transparent";
node[width=0.15, height=0.15, shape=circle, color=white, fontcolor=white];
edge[weight=2, arrowhead=none, color=white];
node[group=main];
1 -> 2 -> 3 -> 4 -> 5;
node[group=branch1];
2 -> 6 -> 7 -> 4;
}
These nodes can be represented (right to left) in topological order as
either 5, 4, 3, 7, 6, 2, 1 or 5, 4, 7, 6, 3, 2, 1.
Having no further children, 5 is a leaf node, and can be used as a
group. All first parents of 5 can therefore be considered to be in
group 5.
7 is a second parent to 4, and so should be used as the group for
all of its parents not present in group 5.
(defun git-graph/group-topo (nodelist)
(reverse
(car
(-reduce-from
(lambda (acc node)
(let* ((grouped-nodes (car acc))
(group-stack (cdr acc))
(node-id (git-graph/node-id node))
(group-from-stack (--if-let (assoc node-id group-stack)
(cdr it)))
(group (or group-from-stack node-id))
(parents (git-graph/node-parents node))
(first-parent (first parents)))
(if group-from-stack
(pop group-stack))
(if (and first-parent (not (assoc first-parent group-stack)))
(push (cons first-parent group) group-stack))
(cons (cons (git-graph/make-node node-id
parents
`((group . ,group)
(label . ,(git-graph/node-label node))))
grouped-nodes)
group-stack)))
nil
nodelist))))While iterating through the node list, I maintained a stack of pairs built from the first parent of the current node, and the current group. To determine the group, the head of the stack is checked to see if it contains a group for the current node id. If it does, that group is used and it is popped off the stack, otherwise the current node id is used.
The following table illustrates how the stack is used to store and assign group relationships as the process iterates through the node list:
| Node | Parents | Group Stack | Group |
|---|---|---|---|
| 5 | (4) | (4 . 5) | 5 |
| 4 | (3 7) | (3 . 5) | 5 |
| 3 | (2) | (2 . 5) | 5 |
| 7 | (6) | (6 . 7) (2 . 5) | 7 |
| 6 | (2) | (2 . 5) | 7 |
| 2 | (1) | (1 . 5) | 5 |
| 1 | 5 |
Graph without automatic grouping
(git-graph/to-graphviz-pretty
"nogroups"
(list (git-graph/make-node 5 '(4) '((label . master)))
(git-graph/make-node 4 '(3 7))
(git-graph/make-node 3 '(2))
(git-graph/make-node 7 '(6) '((label . develop)))
(git-graph/make-node 6 '(2))
(git-graph/make-node 2 '(1))
(git-graph/make-node 1 nil)))
Graph with automatic grouping
(git-graph/to-graphviz-pretty
"autogroups"
(git-graph/group-topo
(list (git-graph/make-node 5 '(4) '((label . master)))
(git-graph/make-node 4 '(3 7))
(git-graph/make-node 3 '(2))
(git-graph/make-node 7 '(6) '((label . develop)))
(git-graph/make-node 6 '(2))
(git-graph/make-node 2 '(1))
(git-graph/make-node 1 nil))))
Graphing a Git Repository
Satisfied that I had all the necessary tools to start graphing real git repositories, I created an example repository to test against.
Creating a Sample Repository
Using the following script, I created a sample repository to test against. I performed the following actions:
- Forked a develop branch from master.
- Forked a feature branch from develop, with two commits.
- Added another commit to develop.
- Forked a second feature branch from develop, with two commits.
- Merged the second feature branch to develop.
- Merged develop to master and tagged it.
mkdir /tmp/test.git
cd /tmp/test.git
git init
touch README
git add README
git commit -m 'initial'
git commit --allow-empty -m 'first'
git checkout -b develop
git commit --allow-empty -m 'second'
git checkout -b feature-1
git commit --allow-empty -m 'feature 1'
git commit --allow-empty -m 'feature 1 again'
git checkout develop
git commit --allow-empty -m 'third'
git checkout -b feature-2
git commit --allow-empty -m 'feature 2'
git commit --allow-empty -m 'feature 2 again'
git checkout develop
git merge --no-ff feature-2
git checkout master
git merge --no-ff develop
git tag -a 1.0 -m '1.0!'Generating a Graph From a Git Branch
The first order of business was to have a way to call out to git and return the results:
(defun git-graph/git-execute (repo-url command &rest args)
(with-temp-buffer
(shell-command (format "git -C \"%s\" %s"
repo-url
(string-join (cons command args)
" "))
t)
(buffer-string)))
Next, I needed to get the list of commits for a branch in topological
order, with a list of parent commits for each. It turns out git
provides exactly that via its rev-list command.
(defun git-graph/git-rev-list (repo-url head)
(-map (lambda (line) (split-string line))
(split-string (git-graph/git-execute
repo-url
"rev-list" "--topo-order" "--parents" head)
"\n" t)))I also wanted to label branch heads wherever possible. To do this, I looked up the revision name from git, discarding it if it was relative to some other named commit.
(defun git-graph/git-label (repo-url rev)
(let ((name (string-trim
(git-graph/git-execute repo-url
"name-rev" "--name-only" rev))))
(unless (s-contains? "~" name)
name)))Generating the graph for a single branch was as simple as iterating over each commit and creating a node for it.
(defun git-graph/git-graphs-head (repo-url head)
(git-graph/group-topo
(-map (lambda (rev-with-parents)
(let* ((rev (car rev-with-parents))
(parents (cdr rev-with-parents))
(label (git-graph/git-label repo-url rev)))
(git-graph/make-node rev parents
`((label . ,label)))))
(git-graph/git-rev-list repo-url head))))
Here's the result of graphing the master branch:
(git-graph/to-graphviz-pretty
"git"
(git-graph/git-graphs-head
"/tmp/test.git"
"master")) <<graph-git-branch()>>
Graphing Multiple Branches
To graph multiple branches, I needed a function for combining histories. To do so, I simply append any nodes I don't already know about in the first history from the second.
(defun git-graph/+ (a b)
(append a
(-remove (lambda (node)
(assoc (git-graph/node-id node) a))
b)))From there, all that remained was to accumulate the branch histories and output the complete graph:
(defun git-graph/git-load (repo-url heads)
(-reduce #'git-graph/+
(-map (lambda (head)
(git-graph/git-graphs-head repo-url head))
heads)))And here's the example repository, graphed in full:
(git-graph/to-graphviz-pretty
"git"
(git-graph/git-load
"/tmp/test.git"
'("master" "feature-1"))) <<graph-git-repo()>>
Things I may add in the future
Limiting Commits to Graph
Running this against repos with any substantial history can make the graph unwieldy. It'd be a good idea to abstract out the commit list fetching, and modify it to support different ways of limiting the history to display.
Ideas would include:
- Specifying commit ranges
- Stopping at a common ancestor to all graphed branches (e.g., using
git-merge-base). - Other git commit limiting options, like searches, showing only merge or non-merge commits, etc.
Collapsing History
Another means of reducing the size of the resulting graph would be to collapse unimportant sections of it. It should be possible to collapse a section of the graph, showing a count of skipped nodes.
The difficult part would be determining what parts aren't worth drawing. Something like this would be handy, though, for concisely graphing the state of multiple ongoing development branches (say, to get a picture of what's been going on since the last release, and what's still incomplete).
digraph G {
rankdir="LR";
bgcolor="transparent";
node[width=0.15,height=0.15,shape=point,color=white];
edge[weight=2,arrowhead=none,color=white];
node[group=main];
1 -> 2 -> 3 -> 4 -> 5;
node[group=branch];
2 -> 6 -> 7 -> 8 -> 9 -> 10 -> 4;
}
digraph G {
rankdir="LR";
bgcolor="transparent";
node[width=0.15,height=0.15,shape=point,color=white];
edge[weight=2,arrowhead=none,color=white,fontcolor=white];
node[group=main];
1 -> 2 -> 3 -> 4 -> 5;
node[group=branch];
2 -> 6;
6 -> 10[style=dashed,label="+3"];
10 -> 4;
}
Clean up and optimize the code a bit
Some parts of this (particularly, the grouping) are probably pretty inefficient. If this turns out to actually be useful, I may take another crack at it.
Final Code
In case anyone would like to use this code for anything, or maybe just pick it apart and play around with it, all the Emacs Lisp code in this post is collected into a single file below:
;;; git-graph.el --- Generate git-style graphs using graphviz
;; Copyright (c) 2015 Correl Roush <correl@gmail.com>
;;; License:
;; This program is free software; you can redistribute it and/or modify
;; it under the terms of the GNU General Public License as published by
;; the Free Software Foundation; either version 3, or (at your option)
;; any later version.
;;
;; This program is distributed in the hope that it will be useful,
;; but WITHOUT ANY WARRANTY; without even the implied warranty of
;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
;; GNU General Public License for more details.
;;
;; You should have received a copy of the GNU General Public License
;; along with GNU Emacs; see the file COPYING. If not, write to the
;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
;; Boston, MA 02110-1301, USA.
;;; Commentary:
;;; Code:
(require 'dash)
<<git-graph/structure>>
<<git-graph/adder>>
<<git-graph/to-graphviz>>
<<git-graph/to-graphviz-nodes>>
<<git-graph/to-graphviz-edges>>
<<git-graph/group-topo>>
<<git-graph/from-git>>
(provide 'git-graph)
;;; git-graph.el ends hereDownload: git-graph.el
Use a different theme when publishing Org files
CLOSED: [2016-02-23 Tue]
I've been using material-theme lately, and I sometimes switch around, but I've found that solarized produces the best exported code block results. To avoid having to remember to switch themes when exporting, I wrote a quick wrapper for org-export to do it for me:
(defun my/with-theme (theme fn &rest args)
(let ((current-themes custom-enabled-themes))
(mapcar #'disable-theme custom-enabled-themes)
(load-theme theme t)
(let ((result (apply fn args)))
(mapcar #'disable-theme custom-enabled-themes)
(mapcar (lambda (theme) (load-theme theme t)) current-themes)
result)))
(advice-add #'org-export-to-file :around (apply-partially #'my/with-theme 'solarized-dark))
(advice-add #'org-export-to-buffer :around (apply-partially #'my/with-theme 'solarized-dark))Voilà, no more bizarrely formatted code block exports from whatever theme I might have loaded at the time :)
Recursive HTTP Requests with Elm
CLOSED: [2018-01-22 Mon]
So I got the idea in my head that I wanted to pull data from the GitLab / GitHub APIs in my Elm app. This seemed straightforward enough; just wire up an HTTP request and a JSON decoder, and off I go. Then I remember, oh crap… like any sensible API with a potentially huge amount of data behind it, the results come back paginated. For anyone unfamiliar, this means that a single API request for a list of, say, repositories, is only going to return up to some maximum number of results. If there are more results available, there will be a reference to additional pages of results, that you can then fetch with another API request. My single request decoding only the results returned from that single request wasn't going to cut it.
I had a handful of problems to solve. I needed to:
- Detect when additional results were available.
- Parse out the URL to use to fetch the next page of results.
- Continue fetching results until none remained.
- Combine all of the results, maintaining their order.
Are there more results?
The first two bullet points can be dealt with by parsing and inspecting the response header. Both GitHub and GitLab embed pagination links in the HTTP Link header. As I'm interested in consuming pages until no further results remain, I'll be looking for a link in the header with the relationship "next". If I find one, I know I need to hit the associated URL to fetch more results. If I don't find one, I'm done!
Link: <https://api.github.com/user/repos?page=3&per_page=100>; rel="next",
<https://api.github.com/user/repos?page=50&per_page=100>; rel="last"Parsing this stuff out went straight into a utility module.
module Paginated.Util exposing (links)
import Dict exposing (Dict)
import Maybe.Extra
import Regex
{-| Parse an HTTP Link header into a dictionary. For example, to look
for a link to additional results in an API response, you could do the
following:
Dict.get "Link" response.headers
|> Maybe.map links
|> Maybe.andThen (Dict.get "next")
-}
links : String -> Dict String String
links s =
let
toTuples xs =
case xs of
[ Just a, Just b ] ->
Just ( b, a )
_ ->
Nothing
in
Regex.find
Regex.All
(Regex.regex "<(.*?)>; rel=\"(.*?)\"")
s
|> List.map .submatches
|> List.map toTuples
|> Maybe.Extra.values
|> Dict.fromList
A little bit of regular expression magic, tuples, and
Maybe.Extra.values to keep the matches, and now I've got my
(Maybe) URL.
Time to make some requests
Now's the time to define some types. I'll need a Request, which will
be similar to a standard Http.Request, with a slight difference.
type alias RequestOptions a =
{ method : String
, headers : List Http.Header
, url : String
, body : Http.Body
, decoder : Decoder a
, timeout : Maybe Time.Time
, withCredentials : Bool
}
type Request a
= Request (RequestOptions a)
What separates it from a basic Http.Request is the decoder field
instead of an expect field. The expect field in an HTTP request is
responsible for parsing the full response into whatever result the
caller wants. For my purposes, I always intend to be hitting a JSON
API returning a list of items, and I have my own designs on parsing
bits of the request to pluck out the headers. Therefore, I expose only
a slot for including a JSON decoder representing the type of item I'll
be getting a collection of.
I'll also need a Response, which will either be Partial
(containing the results from the response, plus a Request for
getting the next batch), or Complete.
type Response a
= Partial (Request a) (List a)
| Complete (List a)
Sending the request isn't too bad. I can just convert my request into
an Http.Request, and use Http.send.
send :
(Result Http.Error (Response a) -> msg)
-> Request a
-> Cmd msg
send resultToMessage request =
Http.send resultToMessage <|
httpRequest request
httpRequest : Request a -> Http.Request (Response a)
httpRequest (Request options) =
Http.request
{ method = options.method
, headers = options.headers
, url = options.url
, body = options.body
, expect = expect options
, timeout = options.timeout
, withCredentials = options.withCredentials
}
expect : RequestOptions a -> Http.Expect (Response a)
expect options =
Http.expectStringResponse (fromResponse options)
All of my special logic for handling the headers, mapping the decoder
over the results, and packing them up into a Response is baked into
my Http.Request via a private fromResponse translator:
fromResponse :
RequestOptions a
-> Http.Response String
-> Result String (Response a)
fromResponse options response =
let
items : Result String (List a)
items =
Json.Decode.decodeString
(Json.Decode.list options.decoder)
response.body
nextPage =
Dict.get "Link" response.headers
|> Maybe.map Paginated.Util.links
|> Maybe.andThen (Dict.get "next")
in
case nextPage of
Nothing ->
Result.map Complete items
Just url ->
Result.map
(Partial (request { options | url = url }))
itemsPutting it together
Now, I can make my API request, and get back a response with
potentially partial results. All that needs to be done now is to make
my request, and iterate on the results I get back in my update
method.
To make things a bit easier, I add a method for concatenating two responses:
update : Response a -> Response a -> Response a
update old new =
case ( old, new ) of
( Complete items, _ ) ->
Complete items
( Partial _ oldItems, Complete newItems ) ->
Complete (oldItems ++ newItems)
( Partial _ oldItems, Partial request newItems ) ->
Partial request (oldItems ++ newItems)Putting it all together, I get a fully functional test app that fetches a paginated list of repositories from GitLab, and renders them when I've fetched them all:
module Example exposing (..)
import Html exposing (Html)
import Http
import Json.Decode exposing (field, string)
import Paginated exposing (Response(..))
type alias Model =
{ repositories : Maybe (Response String) }
type Msg
= GotRepositories (Result Http.Error (Paginated.Response String))
main : Program Never Model Msg
main =
Html.program
{ init = init
, update = update
, view = view
, subscriptions = \_ -> Sub.none
}
init : ( Model, Cmd Msg )
init =
( { repositories = Nothing }
, getRepositories
)
update : Msg -> Model -> ( Model, Cmd Msg )
update msg model =
case msg of
GotRepositories (Ok response) ->
( { model
| repositories =
case model.repositories of
Nothing ->
Just response
Just previous ->
Just (Paginated.update previous response)
}
, case response of
Partial request _ ->
Paginated.send GotRepositories request
Complete _ ->
Cmd.none
)
GotRepositories (Err _) ->
( { model | repositories = Nothing }
, Cmd.none
)
view : Model -> Html Msg
view model =
case model.repositories of
Nothing ->
Html.div [] [ Html.text "Loading" ]
Just (Partial _ _) ->
Html.div [] [ Html.text "Loading..." ]
Just (Complete repos) ->
Html.ul [] <|
List.map
(\x -> Html.li [] [ Html.text x ])
repos
getRepositories : Cmd Msg
getRepositories =
Paginated.send GotRepositories <|
Paginated.get
"http://git.phoenixinquis.net/api/v4/projects?per_page=5"
(field "name" string)There's got to be a better way
I've got it working, and it's working well. However, it's kind of a
pain to use. It's nice that I can play with the results as they come
in by peeking into the Partial structure, but it's a real chore to
have to stitch the results together in my application's update
method. It'd be nice if I could somehow encapsulate that behavior in
my request and not have to worry about the pagination at all in my
app.
It just so happens that, with Tasks, I can.
Feel free to check out the full library documentation and code referenced in this post here.
Continue on with part two, Cleaner Recursive HTTP Requests with Elm Tasks.
Cleaner Recursive HTTP Requests with Elm Tasks
CLOSED: [2018-01-23 Tue]
Continued from part one, Recursive HTTP Requests with Elm.
In my last post, I described my first pass at building a library to fetch data from a paginated JSON REST API. It worked, but it wasn't too clean. In particular, the handling of the multiple pages and concatenation of results was left up to the calling code. Ideally, both of these concerns should be handled by the library, letting the application focus on working with a full result set. Using Elm's Tasks, we can achieve exactly that!
What's a Task?
A Task is a data structure in Elm which represents an asynchronous
operation that may fail, which can be mapped and chained. What this
means is, we can create an action, transform it, and chain it with
additional actions, building up a complex series of things to do into
a single Task, which we can then package up into a Cmd and hand to
the Elm runtime to perform. You can think of it like building up a
Future or Promise, setting up a sort of callback chain of mutations
and follow-up actions to be taken. The Elm runtime will work its way
through the chain and hand your application back the result in the
form of a Msg.
So, tasks sound great!
Moving to Tasks
Just to get things rolling, let's quit using Http.send, and instead
prepare a simple toTask function leveraging the very handy
Http.toTask. This'll give us a place to start building up some more
complex behavior.
send :
(Result Http.Error (Response a) -> msg)
-> Request a
-> Cmd msg
send resultToMessage request =
toTask request
|> Task.attempt resultToMessage
toTask : Request a -> Task Http.Error (Response a)
toTask =
httpRequest >> Http.toTaskShifting the recursion
Now, for the fun bit. We want, when a request completes, to inspect
the result. If the task failed, we do nothing. If it succeeded, we
move on to checking the response. If we have a Complete response,
we're done. If we do not, we want to build another task for the next
request, and start a new iteration on that.
All that needs to be done here is to chain our response handling using
Task.andThen, and either recurse to continue the chain with the next
Task, or wrap up the final results with Task.succeed!
recurse :
Task Http.Error (Response a)
-> Task Http.Error (Response a)
recurse =
Task.andThen
(\response ->
case response of
Partial request _ ->
httpRequest request
|> Http.toTask
|> recurse
Complete _ ->
Task.succeed response
)
That wasn't so bad. The function recursion almost seems like cheating:
I'm able to build up a whole chain of requests based on the results
without actually having the results yet! The Task lets us define a
complete plan for what to do with the results, using what we know
about the data structures flowing through to make decisions and tack
on additional things to do.
Accumulating results
There's just one thing left to do: we're not accumulating results yet. We're just handing off the results of the final request, which isn't too helpful to the caller. We're also still returning our Response structure, which is no longer necessary, since we're not bothering with returning incomplete requests anymore.
Cleaning up the types is pretty easy. It's just a matter of switching
out some instances of Response a with List a in our type
declarations…
send :
(Result Http.Error (List a) -> msg)
-> Request a
-> Cmd msg
toTask : Request a -> Task Http.Error (List a)
recurse :
Task Http.Error (Response a)
-> Task Http.Error (List a)
…then changing our Complete case to return the actual items:
Complete xs ->
Task.succeed xs
The final step, then, is to accumulate the results. Turns out this is
super easy. We already have an update function that combines two
responses, so we can map that over our next request task so that it
incorporates the previous request's results!
Partial request _ ->
httpRequest request
|> Http.toTask
|> Task.map (update response)
|> recurseTidying up
Things are tied up pretty neatly, now! Calling code no longer needs to
care whether the JSON endpoints its calling paginate their results,
they'll receive everything they asked for as though it were a single
request. Implementation details like the Response structure,
update method, and httpRequest no longer need to be exposed.
toTask can be exposed now as a convenience to anyone who wants to
perform further chaining on their calls.
Now that there's a cleaner interface to the module, the example app is looking a lot cleaner now, too:
module Example exposing (..)
import Html exposing (Html)
import Http
import Json.Decode exposing (field, string)
import Paginated
type alias Model =
{ repositories : Maybe (List String) }
type Msg
= GotRepositories (Result Http.Error (List String))
main : Program Never Model Msg
main =
Html.program
{ init = init
, update = update
, view = view
, subscriptions = \_ -> Sub.none
}
init : ( Model, Cmd Msg )
init =
( { repositories = Nothing }
, getRepositories
)
update : Msg -> Model -> ( Model, Cmd Msg )
update msg model =
case msg of
GotRepositories result ->
( { model | repositories = Result.toMaybe result }
, Cmd.none
)
view : Model -> Html Msg
view model =
case model.repositories of
Nothing ->
Html.div [] [ Html.text "Loading" ]
Just repos ->
Html.ul [] <|
List.map
(\x -> Html.li [] [ Html.text x ])
repos
getRepositories : Cmd Msg
getRepositories =
Paginated.send GotRepositories <|
Paginated.get
"http://git.phoenixinquis.net/api/v4/projects?per_page=5"
(field "name" string)
So, there we have it! Feel free to check out the my complete
Paginated library on the Elm package index, or on GitHub. Hopefully
you'll find it or this post useful. I'm still finding my way around
Elm, so any and all feedback is quite welcome :)
How Does The Phillips Hue Wake-Up Feature Work?
CLOSED: [2018-03-13 Tue]
I recently got myself a set of Phillips Hue White and Color Ambiance lights. One of the features I was looking forward to in particular (besides playing with all the color options) was setting a wake-up alarm with the lights gradually brightening. This was pretty painless to get set up using the phone app. I'm pretty happy with the result, but there's certainly some things I wouldn't mind tweaking. For example, the initial brightness of the bulbs (at the lowest setting) still seems a bit bright, so I might want to delay the bedside lamps and let the more distant lamp start fading in first. I also want to see if I can fiddle it into transitioning between some colors to get more of a sunrise effect (perhaps "rising" from the other side of the room, with the light spreading towards the head of the bed).
Figuring out how the wake-up settings that the app installed on my bridge seemed a good first step towards introducing my own customizations.
Information on getting access to a Hue bridge to make REST API calls to it can be found in the Hue API getting started guide.
My wake-up settings
My wake-up is scheduled for 7:00 to gradually brighten the lights with a half-hour fade-in each weekday. I also toggled on the setting to automatically turn the lights off at 9:00.
Finding things on the bridge
The most natural starting point is to check the schedules. Right off the bat, I find what I'm after:
The schedule …
GET http://bridge/api/${username}/schedules/1{
"name": "Wake up",
"description": "L_04_fidlv_start wake up",
"command": {
"address": "/api/xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/sensors/2/state",
"body": {
"flag": true
},
"method": "PUT"
},
"localtime": "W124/T06:30:00",
"time": "W124/T10:30:00",
"created": "2018-03-11T19:46:54",
"status": "enabled",
"recycle": true
}
This is a recurring schedule item that runs every weekday at 6:30. We
can tell this by looking at the localtime field. From the
documentation on time patterns, we can see that it's a recurring time
pattern specifying days of the week as a bitmask, and a time (6:30).
0MTWTFSS |
01111100 (124 in decimal) |
Since this schedule is enabled, we can be assured that it will run,
and in doing so, will issue a PUT to a sensors endpoint, setting a
flag to true.
… triggers the sensor …
GET http://bridge/api/${username}/sensors/2{
"state": {
"flag": false,
"lastupdated": "2018-03-13T13:00:00"
},
"config": {
"on": true,
"reachable": true
},
"name": "Sensor for wakeup",
"type": "CLIPGenericFlag",
"modelid": "WAKEUP",
"manufacturername": "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
"swversion": "A_1801260942",
"uniqueid": "L_04_fidlv",
"recycle": true
}The sensor is what's really setting things in motion. Here we've got a generic CLIP flag sensor that is triggered exclusively by our schedule. Essentially, by updating the flag state, we trigger the sensor.
… triggers a rule …
GET http://bridge/api/${username}/rules/1{
"name": "L_04_fidlv_Start",
"owner": "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
"created": "2018-03-11T19:46:51",
"lasttriggered": "2018-03-13T10:30:00",
"timestriggered": 2,
"status": "enabled",
"recycle": true,
"conditions": [
{
"address": "/sensors/2/state/flag",
"operator": "eq",
"value": "true"
}
],
"actions": [
{
"address": "/groups/1/action",
"method": "PUT",
"body": {
"scene": "7GJer2-5ahGIqz6"
}
},
{
"address": "/schedules/2",
"method": "PUT",
"body": {
"status": "enabled"
}
}
]
}
Now things are happening. Looking at the conditions, we can see that
this rule triggers when the wakeup sensor updates, and its flag is set
to true. When that happens, the bridge will iterate through its
rules, find that the above condition has been met, and iterate through
each of the actions.
… which sets the scene …
The bedroom group (/groups/1 in the rule's action list) is set to
the following scene, which turns on the lights at minimum brightness:
GET http://bridge/api/${username}/scenes/7GJer2-5ahGIqz6{
"name": "Wake Up init",
"lights": [
"2",
"3",
"5"
],
"owner": "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
"recycle": true,
"locked": true,
"appdata": {},
"picture": "",
"lastupdated": "2018-03-11T19:46:50",
"version": 2,
"lightstates": {
"2": {
"on": true,
"bri": 1,
"ct": 447
},
"3": {
"on": true,
"bri": 1,
"ct": 447
},
"5": {
"on": true,
"bri": 1,
"ct": 447
}
}
}… and schedules the transition …
Another schedule (/schedules/2 in the rule's action list) is enabled
by the rule.
GET http://bridge/api/${username}/schedules/2{
"name": "L_04_fidlv",
"description": "L_04_fidlv_trigger end scene",
"command": {
"address": "/api/xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/groups/0/action",
"body": {
"scene": "gXdkB1um68N1sZL"
},
"method": "PUT"
},
"localtime": "PT00:01:00",
"time": "PT00:01:00",
"created": "2018-03-11T19:46:51",
"status": "disabled",
"autodelete": false,
"starttime": "2018-03-13T10:30:00",
"recycle": true
}
This schedule is a bit different from the one we saw before. It is
normally disabled, and it's time pattern (in localtime) is
different. The PT prefix specifies that this is a timer which
expires after the given amount of time has passed. In this case, it is
set to one minute (the first 60 seconds of our wake-up will be spent
in minimal lighting). Enabling this schedule starts up the timer. When
one minute is up, another scene will be set.
This one, strangely, is applied to group 0, the meta-group including
all lights, but since the scene itself specifies to which lights it
applies, there's no real problem with it.
… to a fully lit room …
GET http://bridge/api/${username}/scenes/gXdkB1um68N1sZL{
"name": "Wake Up end",
"lights": [
"2",
"3",
"5"
],
"owner": "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
"recycle": true,
"locked": true,
"appdata": {},
"picture": "",
"lastupdated": "2018-03-11T19:46:51",
"version": 2,
"lightstates": {
"2": {
"on": true,
"bri": 254,
"ct": 447,
"transitiontime": 17400
},
"3": {
"on": true,
"bri": 254,
"ct": 447,
"transitiontime": 17400
},
"5": {
"on": true,
"bri": 254,
"ct": 447,
"transitiontime": 17400
}
}
}
This scene transitions the lights to full brightness over the next 29
minutes (1740 seconds), per the specified transitiontime (which is
specified in deciseconds).
… which will be switched off later.
Finally, an additional rule takes care of turning the lights off and the wake-up sensor at 9:00 (Two and a half hours after the initial triggering of the sensor).
GET http://bridge/api/${username}/rules/2{
"name": "Wake up 1.end",
"owner": "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
"created": "2018-03-11T19:46:51",
"lasttriggered": "2018-03-13T13:00:00",
"timestriggered": 2,
"status": "enabled",
"recycle": true,
"conditions": [
{
"address": "/sensors/2/state/flag",
"operator": "eq",
"value": "true"
},
{
"address": "/sensors/2/state/flag",
"operator": "ddx",
"value": "PT02:30:00"
}
],
"actions": [
{
"address": "/groups/2/action",
"method": "PUT",
"body": {
"on": false
}
},
{
"address": "/sensors/2/state",
"method": "PUT",
"body": {
"flag": false
}
}
]
}
Unlike the first rule, this one doesn't trigger immediately. It has an
additional condition on the sensor state flag using the special ddx
operator, which (given the timer specified) is true two and a half
hours after the flag has been set. As the schedule sets it at 6:30,
that means that this rule will trigger at 9:00, turn the lights off in
the bedroom, and set the sensor's flag to false.
Where to go from here
The wake-up config in the phone app touched on pretty much every major aspect of the Hue bridge API. Given the insight I now have into how it works, I can start constructing my own schedules and transitions, and playing with different ways of triggering them and even having them trigger each other.
If I get around to building my rolling sunrise, I'll be sure to get a post up on it :)
Automating My Apartment With Home Assistant
CLOSED: [2019-06-27 Thu 18:13]
A while ago, I posted about my experiments with the Phillips Hue API to create an automated morning sunrise effect. The end result was nice, but all that mucking about with their HTTP APIs was a hassle any time I wanted to tweak something. I wanted to define what I wanted in a more declarative style, and have all the API calls managed behind the scenes. Home Assistant allowed me to do exactly that, and more.
While the Home Assistant docs are geared heavily towards setting up a raspberry pi appliance to run everything 24/7, I don't own one, and I already have a server going. I opted instead to get the home assistant server running using Docker, and setting up a git repository to hold my configuration.
A Brand New Day
Setting up my sunrise was actually really easy. I already had the scenes I wanted from my previous attempt, so it was just a matter of codifying them in the YAML config. I split them into four scenes - a start (dawn) and end (daylight) pair for the standing lamp at the wall beyond the foot of the bed, and a pair for the two nightstand lights. The end scenes include the transition time to fade in (30 minutes).
scene:
- name: Dawn Sun
entities:
light.standing_lamp:
state: on
brightness: 1
xy_color: [0.6042, 0.3739]
- name: Dawn Daylight
entities:
light.correls_nightstand:
state: on
brightness: 1
xy_color: [0.2376, 0.1186]
light.stephanies_nightstand:
state: on
brightness: 1
xy_color: [0.2376, 0.1186]
- name: Sunrise Sun
entities:
light.standing_lamp:
state: on
transition: 1800
brightness: 254
xy_color: [0.3769, 0.3639]
- name: Sunrise Daylight
entities:
light.correls_nightstand:
state: on
transition: 1800
brightness: 203
xy_color: [0.2698, 0.295]
light.stephanies_nightstand:
state: on
transition: 1800
brightness: 203
xy_color: [0.2698, 0.295]Breaking them apart this way means I can trigger the "sun" first for a splash of orange, then start up the nightstand "daylight" lights a little bit later! This worked out well, too, since even at the lowest brightness, having them turn on right at the start when the room is totally dark had a tendency to jolt me awake. Staggering them produces a much gentler effect. Scripting all of this took very little work…
script:
sunrise:
alias: Sunrise
sequence:
- service: scene.turn_on
data:
entity_id: scene.dawn_sun
- service: scene.turn_on
data:
entity_id: scene.sunrise_sun
- delay:
seconds: 180
- service: scene.turn_on
data:
entity_id: scene.dawn_daylight
- service: scene.turn_on
data:
entity_id: scene.sunrise_daylight… and the end result really is quite pleasant:
That just leaves the automation, which fires a half an hour before the actual sunrise, so long as the lights aren't already on and somebody is home (using a binary sensor I defined elsewhere based on phones detected in the house plus an override toggle).
automation:
- alias: Sunrise
action:
- service: script.sunrise
data: {}
trigger:
- platform: sun
event: sunrise
offset: '-00:30:00'
condition:
- condition: state
entity_id: binary_sensor.occupied
state: 'on'
- condition: state
entity_id: group.bedroom_lights
state: 'off'I later extended the automation with some configuration inputs, which tie into some new triggers and conditions. I added a "latest start time" to make sure it always gets me up in time for me to get ready for work, and an option to disable the wake-up on weekends.
input_select:
sunrise_days:
name: Days to wake up
options:
- Every Day
- Weekdays
initial: Every Day
icon: mdi:weather-sunset
input_datetime:
sunrise_time:
name: Latest start time
has_date: false
has_time: true
initial: '06:30'
automation:
- alias: Sunrise
action:
- service: script.sunrise
data: {}
trigger:
- platform: sun
event: sunrise
offset: '-00:30:00'
- platform: template
value_template: >-
{{ states('sensor.time') == (
states.input_datetime.sunrise_time.attributes.timestamp
| int | timestamp_custom('%H:%M', False)
)
}}
condition:
- condition: state
entity_id: binary_sensor.occupied
state: 'on'
- condition: state
entity_id: group.bedroom_lights
state: 'off'
- condition: or
conditions:
- condition: state
entity_id: input_select.sunrise_days
state: Every Day
- condition: and
conditions:
- condition: state
entity_id: input_select.sunrise_days
state: Weekdays
- condition: time
weekday:
- mon
- tue
- wed
- thu
- friSprinkle in some groups, and I've got a nice panel in my Home Assistant UI to manage everything:
Keep It Down!
Determined to find more things to automate, I realized that since I have my TV audio going through a Sonos sound bar, I could very easily automate the rather annoying ritual of leaping for the app on my phone to turn on night mode when a movie I'm watching is getting explodey and I realize it's a bit late in the evening to be shaking my neighbor's walls.
automation:
- alias: Toggle Sonos night mode
action:
- service: media_player.sonos_set_option
entity_id: media_player.den
data_template:
night_sound: >-
{{ now().hour >= 22 }}
trigger:
- platform: time
at: '22:30:00'
- platform: time
at: '08:00:00'Boom. Happier neighbors, and I can fall asleep in front of movies without worry!
Just because I could, I also added some configurability to this automation as well. The logic got a bit tricky, since I wanted to configure a window that crosses a 24-hour boundary. I also added a binary sensor so I could see when night mode was enabled from Home Assistant.
automation:
- alias: Toggle Sonos night mode
action:
- service: media_player.sonos_set_option
entity_id: media_player.den
data_template:
night_sound: >-
{% set start = states.input_datetime.sonos_nightmode_start.attributes %}
{% set end = states.input_datetime.sonos_nightmode_end.attributes %}
{% set now_ = (now().hour, now().minute, now().second) %}
{% set start_ = (start.hour, start.minute, start.second) %}
{% set end_ = (end.hour, end.minute, end.second) %}
{% if start_ > end_ -%}
{{ now_ >= start_ or now_ < end_ }}
{%- else -%}
{{ now_ >= start_ and now_ < end_ }}
{%- endif -%}
trigger:
- platform: template
value_template: "{{ states('sensor.time') == (states.input_datetime.sonos_nightmode_start.attributes.timestamp | int | timestamp_custom('%H:%M', False)) }}"
- platform: template
value_template: "{{ states('sensor.time') == (states.input_datetime.sonos_nightmode_end.attributes.timestamp | int | timestamp_custom('%H:%M', False)) }}"
sensor:
- platform: time_date
display_options:
- time
input_datetime:
sonos_nightmode_start:
name: Start Night Mode
has_date: false
has_time: true
initial: '22:30'
sonos_nightmode_end:
name: End Night Mode
has_date: false
has_time: true
initial: '08:00'
binary_sensor:
- platform: template
sensors:
den_night_mode:
friendly_name: Sonos Den Night Mode
value_template: >-
{{ state_attr('media_player.den', 'night_sound') }} And, voilà, a dashboard for my speakers, which I pretty much never need to look at anymore!
But Wait, There's More!
It's a too much to cover in a single blog post, but there's plenty more going on in my config. Over time, I've tweaked and added to my device tracking to make sure Home Assistant knows when someone's home. I set up some text-to-speech to announce the weather in the morning, and welcome the first person to get home. I even re-purposed an old phone as a webcam so I can check on the cat while I'm out. My config is on my personal gitlab server, feel free to check it out and see if there's anything there you can use or learn from: http://git.phoenixinquis.net/correlr/home-assistant
Trans Day of Visibility 2022
CLOSED: [2022-03-31 Thu]
It's hard to feel positive about this year's Transgender Day of Visibility. On the one hand, trans visibility is extremely important. It's because of out trans people that I was able to understand my own identity. The more cis people really see, talk to, and come to understand trans people, the easier it will be for them to understand that we're, well, just people. Transitioning is a beautiful thing. Look at any set of photos trans people share, and you'll see that they're not just happier, but more vibrant, more full of life, and so very genuinely themselves! This is what folks need to see more of, and what I think this day is meant to be about. Unfortunately, a lot of what folks are seeing nowadays isn't trans people thriving, it's misinformation and vitriol. This isn't at all a new phenomenon, but in recent years it's gotten overwhelming.
This year, like last year, has brought with it a record-breaking amount of anti-trans legislation across the majority of states in the country. These bills are targeting trans youths by banning them from playing sports with their peers, forbidding any discussion about gender or queer identities in their classrooms, requiring that trusted teachers and other school staff out them to families, and restricting and even outlawing their healthcare. Book bans have been sweeping the nation, intent on removing anything they consider unpleasant or uncomfortable, which has mostly amounted to anything discussing gender, sexuality, or race. There is a constant stream of vitriol flowing across social media and news outlets sowing outrage by recycling old homophobic rhetoric as they label trans people as predators, anyone supporting us as "groomers", and claim we're forcing children into life-altering surgeries. Trans kids do not get surgeries, but laws are being pushed and passed banning them anyway, though always with a note that those restrictions aren't extended to intersex kids, who continue to be operated upon to make their bodies conform to a comfortable binary.
Trans kids and trans adults alike, whether they're in states that are actively arguing or passing these bills, are having to endure watching this all happen. Watching their identities, their existence be debated, questioned, demonized, and ridiculed. We're having to watch this all unfold, and it really feels like few people are actively defending us or standing up to this torrent of hate. Most of these bills aren't even getting much news coverage, and those that are often aren't in our favor, framing the issues as divisive or controversial. Even Florida's so-called "Don't Say Gay" bill is framed first and foremost as an attack on gay rights (which it certainly is), but leaving the very deliberate targeting of trans kids out of the discussion. Florida governor Ron DeSantis certainly didn't hide it, claiming its intent is to squash so-called "woke gender ideology" and pointing at a large illustration from a transgender author's childrens book just before signing the bill.
It's hard, as a trans person watching all of this, seeing these kids, their parents, and ourselves under such constant and cruel attack. It's hard hearing only the faintest murmurs of "we've got your back" from the White House as the Equality Act continues to languish, stalled in Congress. It is hard seeing so few people outside of the transgender community, traumatized so much by it as it is, raising any awareness of what's going on. Each year we endure so much. We watch public figures tell people we're perverts and predators. We watch where we go and what we do in public, lest we inadvertently draw anyone's ire. We watch as some trans folks do succeed, and celebrate them, but also see all the nastiness directed at them in the media and in strangers' comments on social media. All of this is so, so traumatizing.
Ours is a community molded by trauma and loss. Our history, vibrant as it is, has been largely hidden from us or outright destroyed. Nearly an entire generation of queer people was lost to hate and apathy during the AIDS epidemic. Many continue to be lost every year to violence. Mostly trans women of color, losing their lives to hate in the rising tide of racism, misogyny, homophobia and transphobia. We likely lose far more than we know as crimes go unreported or misreported, as they tend to be, when trans folks get misgendered in death. This isn't how it's supposed to be. Discovering and living as who we truly are is one of the most joyful things in life. Being ourselves, really sharing ourselves with the people we love is such a wonderful, vibrant feeling. That more and more people are able to learn about the beautiful spectrums of identities is an amazing thing. We've got greater resources and representation now than ever before.
I do not believe that all of this hatred, all of these laws, any of it will win out in the end. Trans people aren't going anywhere; being trans is just a part of being human. We have always been here, and we will continue to be. What I fear isn't that trans people will be wiped off the earth, because that's impossible. What I do fear is how hard the struggle may remain for us to continue to just live. I feel for these kids, terrified as they are that the world hates them. I feel for the trans community, as we struggle with vastly different degrees of discrimination and violence. It's a lot.
On this Transgender Day of Visibility, I feel it's important that we're not merely seen, but seen fully. I hope that people will see our joy and our strength and our fierce love of authentic life. I also hope that people will see our pain, and find it in themselves to offer not just performative displays of support but real empathy and action. We're out here showing you who we are and what we can be. Please show us who you are and what we mean to you.
And, for the love of everything, please leave Harry Potter in the past.
Types in Python
Why Use Types?
Success Typing
Running Mypy
Specifying Types
Conference Notes: Strangeloop 2018
CLOSED: [2018-10-04 Thu]
Wednesday
Richard Feldman
<2018-09-26 Wed 09:00-09:30>
- Dictionaries as tables
Grant Maki: Facilitating Technology Change Through Cultural Change
<2018-09-26 Wed 09:40-10:10>
- Trying things in teams
- Tech lead impostor syndrome
- Empathizing with fear, skepticism, and choice overload
Very interesting talk from a tech lead position. Grant emphasized building shared experiences, and approaching adoption of new tools and ideas with empathy for the feelings of each team member.
Ravi Chugh: Structured Editing for Elm in Elm
<2018-09-26 Wed 10:40-11:10>
- "Sketch-n-Sketch"
-
Structured code editor "Deuce"
- Shows AST on code
- Apply context-aware transformations to the AST
- Type inspector mode
- Supports holes
- Live feedback for incomplete programs
-
https://hazel.org - Holes for incomplete code (Cyrus Omar)
- Friday Strangeloop talk
Anthony Deschamps: Programming Robots with Elm
<2018-09-26 Wed 11:30-12:00>
- EV3DEV
- Debian distribution for LEGO Mindstorms
- ?
- Robotics is about perception, behavior, and control
-
PLatform.Worker - Application sans view function
Matthew Griffith: Building a Toolkit for Design
<2018-09-26 Wed 13:00-13:30>
-
Legos
- Everything fits together
- Easy to build
- Easy to change
-
CSS
- Dependent on lots of external factors
- Lots of gotchas
- mdgriffith/elm-ui (Website)
-
Palettes
-
Consists of
- Colors
- Spacing
- Typography
- A concise set of style rules (~50 max)
-
- Each property defines its own tests, which can be run in arbitrary environments
- eightshapes :: Design systems
Dillon Kearns: Extending Type Safety Beyond Elm's Borders
<2018-09-26 Wed 13:40-14:10>
- Addressing "avoidable uncertainty"
-
GraphQL :: dillonkearns/elm-graphql
- Generates code from a specification
-
TypeScript :: dillonkearns/elm-typescript-interop
- Generates TypeScript type declarations for elm ports and flags
- Slides & Resources :: https://bit.ly/typeswithoutborders
Sam Row: Complex Animations Done Well
<2018-09-26 Wed 14:40-15:10>
-
Easings
- Tracks starting and target state
- Animation updates will be jarring if the target state changes
-
Springs
- Tracks current position and velocity, and target state
-
Decoupling animation logic from business logic
- Separate models for application state and visual state
Alley Kelly McKnight: Naming Things in Elm
<2018-09-26 Wed 15:20-15:50>
-
Categories for naming rules
- Naming conventions
- Guiding lights
-
What makes a name good?
- Does the name keep us in the process flow?
- Does the name help the reader build an accurate mental model of the code?
- Does the reader know accurately what the thing you're naming is?
- The Mental Model: Our Human Narrative
The idea of using names to allow functions to flow as a narrative is an interesting one, as is maintaining a consistent metaphor throughout.
Tereza Sokol: Demystifying Parsers
<2018-09-26 Wed 16:00-16:30>
- elm/parser
-
Includes support for building useful error messages
problem : String -> Parser a
I've enjoyed using Haskell's Parsec library in the past, as well as
boost::spirit in C++ back when I first dipped my toes into EBNF
grammars. This will be interesting to play with when I can come up
with a reason to 😊.
Thursday
Shaping our children's education in computing
<2018-09-27 Thu 09:10-10:00>
- Project quantum (questions)
- Teals (high schools)
- Csteachers org
- Raspberry Pi foundation
- Education is rife with unintended consequences - be humble, cooperative
Zero Downtime Migrations of Stateful Systems
<2018-09-27 Thu 10:20-11:00>
-
Three different types of migrations
- perceived zero Downtime
- actually zero Downtime
- no migration of state
-
Perceived Zero Downtime
-
Migrating existing records
- constant syncing
- incremental
-
Building confidence
- automated reconciliation and checks
- row counts
- test environment
-
Decouple interactions
- allow reads
- replay writes
-
Migration
- switch writes to queue
- wait for sync to complete
- switch database
- replay queue
-
-
Actual zero Downtime
- load snapshot
- dual writes
- metrics on any issues
-
No migration
- new service built aware of legacy system
- older data expires
- run in tandem
- delegate to legacy system when needed
- test with dual writes
-
monitoring
- unexpected behaviors
- absent expected state
- slow roll-out
Chasing the Stream Processing Utopia
<2018-09-27 Thu 11:20-12:00>
-
Architecture
- Ingestion (various sources)
- Processing (using Apache Samza)
-
Complexity
- Stateless processing (filtering, data enrichment)
-
Stateful processing (aggregation)
- Windowing
- Beam processing model
-
Accuracy
-
Exactly once processing
- Ingestion (de-dupe)
- Pipelined processing (Chandy Lamport)
- Storing results (idempotent writes / de-dupe)
-
-
Scale
-
Scaling ingestion
- Managed by cloud services, or yourself when self-hosting
-
Scaling processing
-
Challenges
- horizontal scaling
- limited to least performant component
-
Typical bottlenecks
-
data access
- accessing adjunct data (databases, services)
- managing temporary state
- CPU
-
-
- Any source (including batch sources, HDFS)
- Any language
-
Anywhere
- Stream processing as a service
-
Awesome tools
-
Schema management
- Schema evolution
- ACL management
- Topic management
- Stream Processing Job Management
- Data Lineage (e.g. Apache Atlas)
-
Monitoring
- Ingestion (e.g. Kafka Monitor)
- Job Progress (e.g. Burrow)
-
-
Rosie Pattern Language
<2018-09-27 Thu 13:30-14:10>
-
Reasoning
- Regular expressions do not scale
- Combing through data to get devops insights
- libraries of named and namespaced expressions
-
RPL syntax looks like a programming language
- reads/diffs like code
- executable unit tests
- macros
- RPL expressions compose
-
RPL is based on parsing expression crammers, which are greedy and possessive
- makes it difficult to be accidentally inefficient
- Allows parsing recursive grammars
- Logstash (Grok) uses pattern libraries
-
Can your 'grep' do this?
-
curl -s www.google.com | rosie grep -o subs net.url_common- Prints all matching URL sub-expressions
-
sed -n 46,49p /var/log/system.log | rosie match all.thingsall.thingsrepresents a disjunction of a variety of things, different pattern matches will be color-coded
-
head -n 1 /var/log/system.log | rosie grep -o jsonpp num.denoted_hex- JSON-structured output
-
-
Formal basis
-
Chomsky hierarchy
- regular - context-free - context-sensitive - recursively enumerable
- RPL (and all PEG grammars) are context-sensitive
-
-
Shares some similarities with regular expressions
- repetition (same syntax, greedy and possessive)
- character sets (simplified syntax, one name or list or range at a time)
- Operations (look-ahead, negation, etc.) (simplified syntax)
- "choice" differs,
/is ordered choice, possessive)
- Ships with a standard library
-
Debugging
-
Trace a mismatch
echo '17:30:4' | rosie trace time.rfc3339
- REPL!
-
-
Uses
- IBM uses rosie in a big-data pipeline
- Mining source code repositories
- NCSU students wrote RPL patterns to extract language features from different languages (including bash!)
- White-listing valid input (protecting against injection attacks per OWASP)
-
Python bindings! (also, c, go), more coming
- Not very "pythonic", currently
-
Resources
Architecting for Active-Active Resiliency in the Cloud
<2018-09-27 Thu 14:30-15:10>
- Data and Service resiliency and durability
-
Basics (Ingredients)
-
Availability (normal operation time / total time)
- mean time between failure (MTBF) / mean time to repair (MTTR)
-
Multiple components in series (
xandy) (A = Ax * Ay)- Availability decreases
-
Multiple components in parallel (multiple services
x) (A = 1 - (1 - Ax)^2- Availability increases
-
Basic Active Architecture
-
Web -> App -> DB
- cheap, simple, fast to market
- manual intervention, low resiliency, high potential rework
-
-
Basic Active-Passive Architecture
- Load balanced app with fail-over nodes / read-replica db
- Hot or cold standby
- decreased downtime, - backup services & data, good balance
- increased costs, complexity, potential for data loss
-
Basic Active-Active Architecture #1
- Multiple instances taking traffic simultaneously
- always on, minimum data loss, zero intervention
- limited tools, most complex, most expensive
-
-
Considerations & trade-offs (Appetizers)
-
Data Replication
-
Active-Active (avoid this)
-
multiple master databases synchronizing
- collisions
- latency
-
-
Active-Passive
-
Multiple instances writing to a single data store
- promote a read replica on failure
- Active-Active at the service level, Active-Passive at the data level
-
-
Easy data replication databases (managed)
- Cloud Spanner
- Cosmos
- Aurora
-
Easy data replication databases (not managed)
- Apache Cassandra
- MySQL
-
-
Considering tenants - what are we failing over?
- Similar to considering restart strategies in OTP supervision trees
- As availability and durability increases, so does complexity and cost
-
-
Resiliency patterns (The Entrees)
- Circuit breaker (stop retrying)
- Exponential back-off (slow down retries)
- Fallback (error response or cache)
- Multiple instances in different availability zones within a region
- Worker queues and streams decouple components
- Patterns can be combined
-
Tools and services (The Desserts)
- Chaos Monkey
- Serverless
It's important to prepare for failure, and balancing availability and consistency is always a balancing act.
For failure, I like to map out services and their dependency as I would an Erlang supervision tree, drawing boundaries around tightly coupled components, and defining the failure strategy for each group.
Towards Language Support for Distributed Systems
<2018-09-27 Thu 15:30-16:10>
Research is very animated! (doesn't look like it from the outside)
- Language support for distributed system = ways that the programming language/compiler can aid in the development of a distributed application
-
Some areas of research
-
Consistency & Programming Models
-
Consistency?
- Specifically programming models that provide some kind of consistency guarantee (e.g, what guarantees do I have for a replica of an object/datum?)
- Sometimes weak consistency is good enough
-
Conflict-free replicated data types (CRDTs)
- Lasp: A Language for Distributed, Coordination-Free Programming
- Spry is a programming model for building applications that want to trade-off availability and consistency at varying points in application code to support application requirements (CA in CAP)
-
Mixed consistency via "Correctables" (Incremental Consistency Guarantees for Replicated Objects)
-
Give the user different APIs depending on the sort of consistency they're after
invokeWeak(operation)invokeStrong(operation)invoke(operation[, levels])
-
-
Mixed consistency via "MixT" (MixT: A Language for Mixing Consistency in Geodistributed Transactions)
- Answers the question of what if we want to choice from multiple consistency options in one application
-
-
Session Types
- Can be though of as types for communication protocols
-
Many variants of session types
- Binary & Multiparty (# of parties)
- Static & Dynamic
- Introduction to session types by Simon Fowler
-
Prototypes of session types are implemented for many languages
- Erlang
- monitored-session-erlang
- Python
- SPY
- Haskell
- effect-sessions
- Scala
- Ichannels
- Java
- Scribble
- Static Analysis & Verification
-
Others
- Whip (Higher-Order Contracts for Microservices)
-
- Haskell-like language for moving computations around
- Syndicate (Conversational Concurrency)
- ML5 (Modal Types for Mobile Code)
- Function-Passing (A Programming Model and Foundation for Lineage-Based Distributed Computation)
-
Birds-eye view of fringe projects
-
Verifying Interfaces Between Container-Based Components (Or… A Type System By Any Other Name)
- Rejected 😅
- Statically ensuring that microservices satisfy invariants - Adelbert Chang
- Statically ensuring that functions on replicated data are monotone - Kevin Clancy
- Fault-Tolerant Functional Reactive Programming (Functional Pearl)
-
-
-
Resources
Keynote: Erica Joy Baker - Changing the World
<2018-09-27 Thu 16:30-17:20>
- Passionate about talking about diversity and inclusion
- It seems that everyone wants to "change the world"
- "We did it! … Mission accomplished, in the words of George Bush. We changed the world"
- "We got so focused on disruption that we didn't stop to think about who or what we were disrupting"
- Recommends watching "Halt and Catch Fire"
-
"Silicon Valley is Venture Capital"
- Story of "The Scorpion and the Frog"
- The VCs are the scorpions "in case that wasn't clear".
-
Steps to change the world
- Have an idea
- Grow your ideas slowly.
- Get customers
- Get some more of them
- Keep going
- Build what you want to build on your own terms.
- Mailchimp is a sponsor (bleh), has no VC funding
- We have to ask ourselves, "I can do this, but should I do this?"
- "How we're going to change the world is by the legacy we leave behind."
- "We must build companies that represent everyone, and all experiences."
- Call to action: If you work in a company that you know is doing something harmful, that you have averted your eyes from, now is the time to speak up and do something.
- Change the world, just make sure you're changing it for the better.
Friday
All the Languages Together
<2018-09-28 Fri 09:10-10:00>
- Different languages have different strengths
-
Multi-Language software is hard
- FFI/JNI
- microservices
-
Can we allow reasoning about interop at the source level?
- As opposed to running code and encountering errors / debugging
-
Language specifications are incomplete - don't account for linking
- 'escape hatches' (ML/C FFI, Rust/unsafe, Java/JNI)
- We tell programmers to "be careful"
-
Rethink PL Design: Principled FFIs
-
Design linking types extensions that support safe interoperability with other languages
- linear types (similar to rust ownership)
- pure (terminating) functions
- Only need linking type extensions to interact with behavior/features inexpressible in your language
- Linking Types for Multi-Language Software: Have Your Cake and Eat It Too
-
Reasoning about refactoring
- Let the programmer specify, do I want to only link with pure code? Allow linking with stateful code?
-
Minimal annotation burden
- Provide a default translation (e.g, a pure language defaults to linking pure code, not stateful)
-
-
Catching cross-language type errors
- What happens if the linked code doesn't have linking type annotations?
-
Type-preserving compilation
-
Provide specifications of code invariants that can be checked at link time
-
-
Cross-language type errors
- Fail linking if target (compiled) types don't match
-
Refactoring-Preserving Compilation
- Also known as equivalence-preserving compilation, fully abstract compilation, and secure compilation
- Specifying a non-default linking type means giving up some refactoring patterns
- Programmer gets to decide what refactorings matter to them; LT annotations tell the compiler what linking to allow/disallow
- Linking types are the escape hatch
-
Ongoing work
-
Mulberry Project
- Building linking types to interop ML with Rust
- Building linking types for a pure language
- Building on top of web assembly (generating a richly typed Wasm with state/effect encapsulation)
-
What about untyped languages (Scheme, C)?
- Untyped Wasm (gradual typing)
- control linking type
-
- SILC - Secure Interoperability, Languages & Compilers
Leverage vs. Autonomy in a Large Software System
<2018-09-28 Fri 10:20-11:00>
-
Big company problems
- How do you make good use of hundreds of software engineers to build a large system?
- How do you organize a large software system so hundreds of engineers working on it can avoid wasting their time?
-
Andy grove on org structure (High Output Management)
- functional team - increased leverage
-
mission-oriented team - increased speed
- speed is the only advantage
-
Conway's Law
- Software tends to look like the organization that built it
-
Coplein's Law
- If your software doesn't look like your organization, you've got trouble
-
Leverage vs Autonomy
- centralized systems - increased leverage
- decentralized systems - increased autonomy
- poor leverage - waste time reinventing wheels
- poor autonomy - waste time waiting on other teams
-
Costs of autonomy/decentralization
- difficult to make global improvements
- difficult to make cross-cutting changes
- uncontrolled system complexity (little consistency)
-
Costs of leverage/centralization
- single points of failure / monoculture risk
- arcade knowledge (downside of knowledge specialization)
- uncontrolled system complexity (poor modularity)
-
sweet spot: cluster scheduling
- increased leverage, removed repetitive operational work
- most customers want roughly the same thing
- no important loss of autonomy
- global system improvement (upgrades)
- bugs (very rarely) bring down multiple services
-
sweet spot: finagle as theft service framework
- high leverage
- most customers want roughly the same thing
- no important loss of autonomy
- difficult to debug
-
sour patch: monolithic service
- lots of autonomy problems (major development choke-point)
- poor sense of ownership
- very easy to cause cross-cutting bugs
-
single ci/deploy pipeline
- easy to make cross-cutting changes
- good leverage
-
sweet spot: service oriented architecture
- autonomy is really good
- every project needs its own ci/deployment pipeline
- difficult to cause cross-cutting bugs
- good ownership
- code is very modularized
- leverage is low
- difficult to make cross-cutting changes
-
sour patch: cross-service application logic library
- where to put cross-cutting logic? attempted a shared library, embedded in a variety of separate services
- low autonomy (lots of services to deploy with changes)
- bad ownership (which team owns this?)
-
sour patch: unowned services
-
consequence of autonomy
- services get lost in the shuffle between teams
-
-
sour patch: ad-hoc service platform
-
building stuff into existing services, rather than creating a new one
- caused by an aversion to build a whole new service
- "What's the fastest way I can get this piece of work done?"
- owners of the code get a lot of leverage, but lose a lot of autonomy
-
-
Strato: platform for microservices
- monolithic service that hosts services inside of it
- captures some common patterns (e.g., a data store with a cache in front of it, etc.)
- Goal is to split the systems concerns from the applications themselves (e.g., timeouts, retries, etc.)
- Microservices within are isolated, allows modularity
- a lot easier for the code structure to match the organization structure
-
Thrift RPC
- communication between services is structurally typed
- "compatibility" relation between types makes upgrades easier
-
StratoQL: a DSL for microservices
- structurally typed
- native support for thrift data
- transparent concurrency (cf. Haxl (previous Sl talk))
-
large gain in leverage over separate microservices
- captures common infrastructure patterns in configuration
- autonomy for application logic owners
- many customers want roughly the same thing
- complicated interface
- centralized team a bottleneck
- difficult to debug
- downtime affects many microservices
Conference Notes: Abstractions II
Day One
Kill All Mutants
- Speaker
- Dave Aronson
- Slides
- https://bit.ly/kill-mutants-abstractions-2019
-
Mutation testing
- mutates code with the goal of generating test failures
- checks for meaningful code and strict tests
- each change to our code should make at least one test fail
- at least one unit test failing on a mutation is "killing the mutant"
- First proposed in Richard Lipton's "Fault Diagnoses of computer programs" (1971)
- Difficult to interpret results, and very CPU intensive
- Python - cosmic-ray, mutmut, xmutant
- JS - stryker
Maintaining a Legacy Haskell App as Not-Yet-Experts
- Speaker
- Libby Horacek
- Slides
- https://slides.com/emhoracek/haskell-24
- Migrated from a haskell app in in frames within wordpress to a haskell app serving wordpress content via the wordpress api
- The "Genius-Oh-No Cycle"
-
Chesterton's Fence
- Understand why a thing is the way it is before you try to change it
- ~3% of functional programmers are women
-
Things that helped
- Pair programming
- TDD (invert the cycle)
-
RFCs
- Can be totally casual
- Sketch out a larger idea and ask for input from the team
Measuring the Human Impact of Software Best Practices: A Story of CSS and Empathy
- Speaker
- jnf
- @_jnf
- Slides
- https://speakerdeck.com/jnf/measuring-the-human-impact-of-software-best-practices-a-story-of-css-and-empathy
- "The Tyranny of Pay-as-you-go Internet"
- Worked at Mighty Ai (training data as a service), acquired by Uber
- Stuff in the talk happened before Uber
-
microaggressions and microaffirmations
- "you guys" vs. (friends, folks, participants, peeps, illustrious heroes, party people, yinz)
-
microaffirmations, a guide
- https://brown.edu/sheridan/microaggressions-and-micro-affirmations-0
- active listening
- recognizing and validating experiences
- affirming emotional reactions
-
micro-tasks in a webapp to generate vector images
- $0.03 to $0.05 USD
- Large base in Venezuela
-
community members raised concerns about data transfer
- forgot ota internet was paid by the megabyte
- didn't know the ost per megabyte
- didn't know there was such flux in data access costs
- asked them to take a literal loss in order to participate in the beta test
- I assumed an experience largely consistent with my own
-
CSS is a tax
- 1.3MB of css vs 34KB after refactoring
- Movistar pricing: 0 MB included, each MB is $5 (5,00 Bs (bolivars))
- 92 deployments, only 9 changed the CSS, every one of them re-generated the bundle and busted the cache
- 8 downloads @ 1.3MB (52Bs) instead of 2 at 34KB (0.34Bs) just for CSS
- This is an ethical issue
- ask, listen, believe, then act
“Testing” - You Keep Using that Word, but I Don’t Think It Means What You Think It Means!
- Speaker
- Samuel Brown
- @SamuelBrownIV
- Slides
- https://bit.ly/testabstract
- Testing is generally re-defined by every organization to fit their culture and needs
- Easy to explain why an acceptance test works, harder to explain what the utility is that you get out of a unit test
-
An opinionated view of testing
- automated testing is the single biggest factor in determining how fast you can ship code
- You have to build up tests in layers and spend time in the right places
- Teams that test write code with better structure focused on interfaces and low cyclomatic complexity
-
Pit of Despair - AKA "Test Environment"
- Don't want to be dependent on the state of the environment to test that your code works
- Useful for UI/UX exploratory testing, load testing, deployment testing
-
What can we do
- Unit tests
- Test the smallest units at the function/method level
- Integration/Component tests
- testing the composition of two functional units or external dependencies that achieve a larger operational function (reduce the number of variables, not increase)
- E2E/Acceptance/UI/API Tests
- Testing features with all required components integrated together but some can be mocked
-
Test Pyramid
- low cost, fast run vs high cost, slow run
- automated unit, automated integration, automated ui/api, manual ui/api
-
Fun tests
- Writing good tests can be as challenging as writing good code
- You will write more (2-3x) test code than feature code
- Testing can be cathartic
-
Strategies for testing
- Legacy code by Michael Feathers
- Consider frameworks and libraries that lend themselves well to testing
- Findexamples for what you want to do
- Keep functions small and purpose-built - If it is hard to test, it's probably too big!
- Limit conditional logic in a single function to 3 levels at most
- Test ALL error conditions respond as expected
- Use stubs, mcks and test doubles to simulate expected inputs and outputs
- Limit integration/component tests to two actors (mocks for all others)
- Create interfaces for components so that they can be mocked
- Store test-case data with your tests
- It's NEVER too late to start writing tests
-
Lightning Sand - AKA Microservices
- The idea is to be (and stay) loosely coupled to the rest of the architecture, minimizing dependencies when testing is key!
-
Strategies for Microservice Testing
- Don't be QA for services you depend on - trust their interface
- Use service mocks for external dependencies
- Write testing libraries/harnesses that other teams can use
- Automate all of your tests - Without a UI you only need robots
-
What about non-functional tests
- Every organization needs to evaluate whether that testing is necessary and correct for them ($$$)
- Collect detailed metrics at service boundaries
- Abstract out service-level concerns like back-pressure, retries and complex routing
How Games Can Inspire Great Application UX
- Speaker
- Scott Jenson
- @scottjenson
- scott@jenson.org
- NOT gamification (don't be on the addictive side of things)
- Video games like the word "juicy" (visual impact on input)
- Video games create tension, apps remove it
-
Ralph Koster (Book: A Theory of Fun)
- Applying the lessons in the book to UX
-
Games present story, players create narrative
- Standalone features vs a journey
-
The Mac story arc
- Sound
- promise
- model
- depth
- Our story isn't the user's narrative
-
Games are fractal, not linear
- Games are made of games
- Each level has its own motivation, feedback, and learning
- The deeper you go, the more profound the effect
-
The Learning Loop
- Intent -> Action -> Result
- Mental model -> affordance -> feedback
- "Fun is just another word for learning" - Ralph
-
Mario 1-1
- move (jump over) -> opening (jump into bricks) -> attacking (jump onto an enemie)
- "That's how we make games at Nintend: we get the fundamentals solid first, then do as much with that core concept as our time and ambition will allow. As forthe courses and enemies, those actually came at the very end, they were done in a single burst of energy…" - Shigeru Miyamoto
- Nintendo does this all the time
- "Desktop has much better loops than mobile" - Ralph
-
Affordances
- BotW stamina wheel - training you to take the better path
- The SonicFinder, An Interface That Uses Auditory Icons
-
Mobile import is impoverished
- Pinch
- Tap
- Long press is a hack!
- Can't find any examples of gracenotes in apps
- "It's not the flashy trick that matters, but the rigourous application of multiple types of feedback in both important and trivial ways" - Ralph
-
Hintiness
- Hints
/Affordances - Affordances reinforce the loop you're on
- Hints move you to a new loop
- Navigation (Disney weenies, always seeing a central place in a park, ensuring sight lines)
- Hintiness prevents "bottom feeding" (getting stuck at one level, not making progress and doing new things)
- Simple, light, patient examples showing how things can be done
- Hints
-
Pacing
- Games work incredibly hard on the first step
- Here is a toolbox vs learn this first
Game Development in Eight Bits
- Speaker
- Kevin Zurawel
- Website
- https://famicom.party
- Slides
- https://bit.ly/gd8b-abs2
- 256x240 resolution with 64 colors
- 1 background layer, 1 sprite layer
-
Backgrounds
- 30x30 grid of tiles, 1 byte per tile
- 2 pattern tables of tiles, 256 tiles each (one for sprites, one for backgrounds)
-
64 colors. 8 of them are black (blame NTSC)
- 8 4-color palettes, 4 for sprites, 4 for backgrounds
- The first color of all pallets is the same (hardware limitation)
-
Sprites
- 256 bytes of sprite ram, 4 bytes per sprite, 64 sprites at a time
-
No more than 8 sprites per scanline
- solved using flickering
-
Level data
- Make use of default color
- abstract elements (pipe, height x)
- run-length encoding
- "set decoration" - three-screens-wide default background (SMB)
-
Physics
- Don't use physics (simple algorithm)
-
Collision detection
-
Contra uses point vs rectangle detection
- The player is always the point, where it can be hit it is a rectangle in relation to the point
-
-
The NES does not have a random number generator (3 optionsin increasingorder of stupidity)
- Tetris: Do it with math (16-bit fibonacci linear feedback shift register)
- FF: (Nasir Gebelli, contractor) a lookup table of 256 random numbers
-
Contra: a single global 8-bit value that increments by 7 whenever the game is idle
- The demo uses prerecorded actions, it can play out differently
-
Saving progress
-
Password systems
- DQ2 in Japan used a "poem"
-
FDS
- Shut down due to ease of piracy
-
Battery-backed memory
- "Hold reset" - power issues could lead to corruption
- Write multiple times with CRC
-
-
"Embrace the stupid"
- Is it close enough, and much more efficient?
Day Two
Duolingo: Microservice Journey
- Speaker
- Max Blaze
- first microservice in 2016
- making many changes to the product, many releases per day
- centralized dashboards/logging
- Terraform for infrastructure as code
- First microservice in ECS in 2017-2018
-
Why move to microservices?
-
Scalability problem with teams
- Slow and difficult with a monolith
- Desire to use multiple languages (monolith in python, wanting to incorporate scala, nodejs, …)
- Flexibility
- Velocity
- Reliability
- Cost savings
-
-
What to carveout first?
- Not the largest chunk
- Start with a small but impactful feature
- move up in size, complexity, and risk
- consider dependencies
- First thing was the reminder service 🦉🗡
- Using circuit breakers to make microservices independent
-
Why docker?
- Kind of the only game in town
-
Why docker with ECS?
- task auto scaling
-
task-level IAM
- needs to be supported by the aws client library (e.g., boto)
- cloudwatch metrics
- dynamic alb targets
- manageability
-
Microservice abstractions at Duolingo
-
Abstracted into terraform modules
-
Web service (internal or external)
- load balancer and route 53
-
worker service (daemon or cron)
- sqs and event-based scheduling
- data store
- monitoring
-
-
-
CI/CD
- Github -> Jenkins -> ECR/Terraform (S3) -> ECS
-
Load balancing
-
ALB vs. CLBs
- ALBs more strict when handling malformed requests (defaults to HTTP/2 (headers always passed in lowercase)
- Differences in cloudwatch metrics (continuous in CLBs, discrete in ALBs)
-
-
Standardizing microservices
- develop a common naming scheme for repos and services
- autogenerate as much of the initial service as possible (?)
- move core functionality to shared base libraries
- provide standard alarms and dashboards
- periodically review microservices for consistency and quality
-
Monitoring microservices
- includes load balancer errors
-
pagerduty integration
- includes links to playbooks
- emergency pages, warnings go to email
- schedules and rotations are managed by terraform
- Grading microservices
-
Cost reduction options
-
Cluster
- instance type
- pricing options
- auto scale
- add/remove AZs
-
using "Spot" (spotinst) to save money on ephermeral cluster instances
- drains ECS services
- spreads capacity across AZs
- bills on % of savings
- ECS allows oversubscription of memory, WE DO NOT RECOMMEND THIS
-
AWS Limits
- EC2 has a hard-coded maximum # of packets(1024/s) sent to an amazon provided dns server
- Nitro is not caching DNS requests where Xen was
-
Mentoring the way to a diverse and inclusive workplace
- Speaker
- Alexandra Millatmal
- @halfghaninNE
- hello@alexandramillatmal.com
- Slides
- http://alexandramillatmal.com/talks
- Developer at Newsela (Education tech company promoting literacy in the 2nd-12th grade space)
- The tenants of mentorship are similar to the tenants of inclusive companies
- Mentorship doesn't work for folks of under-represented backgrounds
- Finding very similar entry level jobs, very homogenous teams, no time to support learning
- Skill-building and diversity appear related
- What if strong mentorship begets diversity & inclusion?
-
Good mentorship / good diversity
-
Supporting and retaining engineers with underrepresented identities
- be welcoming and inclusive in the recruiting process
-
post openings on "key values"
- company must list their values
- conveys people and tech values on the same level
- candidates filter jobs based on their prioritized values
-
referrals can cut both ways
- can increase the homogenous nature of the workplace
- maybe direct referall bonuses into donations to inclusive groups that attract and talent
-
affinity groups
- caucuses across departments working with management
-
standardized review process
-
stanford research into review proceseses
- men overrepresented in the higher tier, women in the middle
- standardizing removed the bias
-
-
clear definitions of roles and responsibilities
- do they have ownership
- are these employees getting a seat at the table for decisions
-
representation in leadership
- are there people there that look like me?
- is there a clear model of advancement? allyship in leadership?
-
investment in internal & external advocacy
- signals that copmanies understand the systematic barriers to inclusion and diversity
- sponsorship - "super mentorship"
- stark differences in valuation of the above bulletpoints between underrepresented groups and well-represented groups (women vs men, lgbt+ vs straight men)
-
-
Supporting and leveling up junior engineers
-
recruiting process / relationships
-
the candidate should be receiving feedback on their performance in the recruting process!
- Gives them constructive advice and context
-
-
apprenticeships and clearly defined entry-level positions
- is there a clear path for growth?
-
clear and structured onboarding
- please do not make their point person a person they report to
- need to go to information from somone that doesn't involve company politics
- information should exist outside of leads/managers heads
- define onboarding procedures in a shared space
-
learning groups
- space to ask questions and demonstrate leadership, particularly with peer-to-peer learning
-
formalized mentorship
- ensure that compensated time is resulting in measurable goals for the junior engineer
- recommend them for opportunities
-
standardized review process
- reframe junior-ness as an opportunity, not a deficit of skill
-
-
Mentorship with diversity and inclusion in mind
-
this work is really hard
- easy to fall into a pattern of saying you're making progress without measuring to make sure that's the case
- intent is only half of the picture
- the other half is sacrifice to make real, measured investments
-
mentorship should begin during the interviews
-
place serious focus on developing mentors
- forces mentees to manage up
- mentorship is a two-way street
- have you ever seen someone become a better collaborator after mentoring a junior engineer?
- mentorship is leadership and it's learned
- have clear growth objectives for the mentor and the mentee
- mentorship should happen on compensated time
-
rethink the peer group
-
slack channel for juniors spread across different offices
- wasn't an organic space to share knowledge
- a black junior woman engineer's peers aren't just other black employees, or women, or other limited groups
-
-
-
What's the value to the company?
- make a business case for mentorship
- that will drive diversity and inclusion
-
mentorship can
- build brand halo among candidates
- distribute management responsibilities
- build its own workforce
- distributes business knowledge working on real business projects
-
fosters relationship building and belonging
- practices wielding expertise, fosters bonding over work
Sextech: The good, the Bad & the Bias
- Speaker
- Alison Falk
- alisonfalkpgh
- Slides
- https://bit.ly/MillatmalAbstractionsII
-
Deepfakes
- people onforums requesting deepfakes of coworkers, etc.
- takes few photos
-
women are most typically targeted
- silenced, made less credible due to the video
- no criminal recourse
- currently thriving
-
nonconsensual / revenge porn
- laws are vague (interstate implications)
-
sexting & development of sexual identity not reflected in the law
- this gets caught under the net of sharing child pornography
-
Just because you can doesn't mean you should
- harms sex workers
- sex trafficing is only 19% of human trafficing
-
25% of sex workers sexually assaulted by officers
- multiple times during stings
- arrests inflate statistics
- 90.8% of victims are deported
-
Bias
- don't need to reach orgasm to procreate? considered a vice, not family planning
-
approved / not approved.com
- approved vs non approved ads
- silencing of minority / repressed groups
-
payment processors
- sex industry is the first adopter of new tech
- kicked off, considered a liability
-
shadowbanning
- facebook's recent patented content filtering algorithm
-
SESTA/FOSTA
- doesn't punish traffickers, makes websites legally liable for any user generated content found to "knowingly assist, facilitate, or support sex trafficing"
- interferes with sex education
- pushes sex trafficking further underground
- based on moral panic (National Center on Sexual Exploitation, formerly Morality in the Media)
- hurts the most vulnerable in our community
-
Root of the problem
- education <=> laws -> media/tech
- who receives accurate sex education
- searches on sex tech aren't showing any of these issues
- teach children to know about their bodies so they don't allow others to make decisions about their bodies
-
sextechspace
- resources online
-
What can you do today
- support orgs like @decrimNY, @decrimNowDC, @TheBADASS_army
- use your platform to spread awareness
- advocate for agency and consent
- make sure all stakeholders are at the table
- "If you are not intentional about being inclusive, what you will do is perpetuate exclusion"
Passing the Torch Without Dropping The Ball: Lessons in Documentation
- Speaker
- Mary Thengvall, Rain Leander
- @mary_grace, @rainleander
- TL;DR: It's not just a question of clocking out and handing over the keys
- Advocato shirts!
- Getting sick, switching roles, moving on
-
Why should you care about a transition plan?
- Taking time off without worrying
- Transitioning into a promotion
- Delegate: documented things allow other people to volunteer
- Are you going to do that forever? document it and let someone else take over
-
The handover document
-
the overview
- project goals
- reading list
-
dramatis personae
- who are the stakeholders/elders?
- learn from them, and document!
- who's the quiet person that gets stuff done?
- who needs a bit more time to ramp up?
- knowledge needs to be documented and disseminated with care, perhaps a bit private
- not toxic, just necessary
- the regular tasks
-
wish list
- that which remains undone
-
the inventory
- budget
- credentials
- the keys to the castle
-
-
eDit edItEdit- what's no longer accurate?
- what's missing?
- what's confusing?
- revisit it, keep it fresh
-
Prioritize
- what's urgent?
- what's important?
- what can only you do?
- what's the low-hanging fruit that's easy to ramp up on?
-
What works for you?
- don't be scared to make changes
- or to say no
- be sure the community agrees with your changes
- Share your stories
The Times They Are a-Changin': A Data-Driven Portrait of New Trends in How We Build Software, Open Source, & What Even is "Entry-Level" Now
- Speaker
- Heather Miller
- @heathercmiller
- Tracking the shift in focus in open source Scala
-
Things that are changing fast
- how we build software
- open source
-
our idea of software engineers
- what should they know?
-
How people are getting into tech
- hiring is difficult
- there's a massive gap between jobs available and people to fill them
- a large portion of professional developers are new
-
we need to adapt, culturally, to make room for lots more newcomers
- frameworks and reuse to reduce friction?
-
existing devs are burning out
- "With companies unable to fill open positions, current employees are expected to fill in the gaps"
-
Increased diversity would help
- Also immigration, remote workers
-
Paper: Gender and tenure diversity in github teams
- increased diversity = increased productivity
-
How do we stop people from disengaging?
- Women disengage earlier than men
-
Open source adoption
- Dramatically increased since 2010
- Open source became the default choice
- Low-cost with no vendor lock-in
- Open source components exist in 96% of applications scanned and analyzed by Synopsys in 2018, with an average of 257 components per application. 36% of code bases were open source components in 2017, 57% in 2018.
- OSS projects 62% self-funded, 49% employer-funded
-
"Truck factor"
- 64% of top projects on Github relied on 1-2 devs to survive
-
Ecosystem and community are everything
- Leo Meyrovich - Paper on programming language adoption
- Most important factor is an ecosystem / open-source libraries
- Professional developers want an active community
- All these puzzle pieces need polishing
Day Three
Analysis of the Feeling of the Attendees to a Talk in Real Time
- Speaker
- Juantomás Garcia Molina
- juantomas
- Google cloud functions triggered by Google Storage, Pub/Sub, or REST
- Using google vision API to detect facial featuers and emotional responses
Identity Expression: Coming Out of the Work Closet
- Speaker
- Matthew Rogers
Coming out of the Closet
- Coming out doesn't just happen once, it happens a lot
-
"Spheres of existence" you have to come out to
- Friends
- Peers
- Family
- Digital
- Public
- Work
-
Story
- The comment is made
- Affected how I work, which affects my coworkers
- Negative effect spreads
- Expands further while training/interacting
- It isn't always cut and dry
- The fear of "what if" can be damaging all its own
-
In 30 states, LGBTQ+ workers aren't fully protected under the law
- PA included
-
Professinal risks
- Fired outright
- Passed over for promotions or projects
- Refusal of reference for next job
-
Personal risks
- Uncomfortable or dangerous work environment
- lost source of income
- Forced to change field or location
-
Ther's risk to the business witself when employees hide part of their identity
-
Productivity
-
Say that 10% of your day goes towards identity concealment
- adds up to 6 weeks of lost productivity every year
-
-
Communication
- less likely to engage
- avoid people
- crosses levels
-
Creativity
- stress + anxiety
- requires vulnerability
-
Collaboration
- needs creativity + communication
- not getting the best work
-
-
What if things went the other way?
- Productivity
- fewer distractions from your work
- communication
- problems caught, efficiency goes up
- creativity
- flourishes in safe + comfortable environments
- collaboration
- becomes easier and more routine
-
Why focus on such a small group of people
- ~4.5% of the entire population
-
Just a method of expressing identity, not limited to just queer people
- 4.5% queer
- 6% practicing non-christian faiths
- 19% have a disability
- 22% persons of color
- 47% female
-
Isn't this just about feelings?
- this holds personal and professional importance with real mental healthy implications
-
Your "Selves"
- Private
- Home
-
You
- Your core self
- Public
- Work
-
Dissonance
- Your genuine self suffers as you put your energy into maintaining a separate self or concealing part of your identity
-
Personal Consequences
- drives down those four factors
- is mentally exhausting
- it becomes a cycle
-
Business consequences
- Quality of Product
- Employee Satisfaction
- Company Culture
- Bottom Line
-
What can I do now?
-
Introduce yourself using your pronouns
- Says you care how other people want to be addressed
-
Start ERGs (Employee resource groups)
- People with shared experiences
- Acknowledge and celebrate Black History Month, Women's Month, Pride Month, etc. Let people feel seen
- Don't ask to touch anyone's hair. Ever.
- Accommodate variations in holidays and scheduling around religious practices
- Don't get embarrassed if you mess up. Apologize, correct yourself, and learn.
-
Replace words like "wife/husband/boyfriend/girlfriend/etc" with partner
- Be careful asking about personal and private relationships in general
- Look into unconscious bias training. Use HR.
-
Look around the room
- Different perspectives and experiences simply work better
- When you don't know, research. Ask questions, if you must. Just don't assume.
-
- If you are in the closet at work, open the door a little
- Make sure you're not blocking someone else's exit
Overcoming Challenges: An Attitude of Iteration
- Speaker
- Eric Johnson
-
"The Look" (first date story)
- If you're different and it shows, it's assumed every problem in the book belongs to you
- "You're not good enough for her" or "That's the saddest thing I've ever seen"
- (context: one finger on each hand, one toe on each foot, 4/5 kids share this trait)
- "Man, I'm different"
- "Things are gonna be a little challenging for me, aren't they?"
- We are all challenged
- How you deal with that challenge is the model by which everyone else is going to deal with it
- 50/50 chance of the genetic condition being passed on (5 fingers, or something else)
-
You have to make the choice of how to deal with it
- Can't force children to make a decision, they have to make it on their own
- It is a blessing, not a curse
- It is what it is
- It'll never be what it isn't
-
No excuses
- "I can't do x yet", not "I can't do x"
- We have to adapt
- Never tell kids they "can't do x",but "it's gonna be hard, you're gonna have to figure it out"
- Have fun
- Sometimes life is hard, but you've still gotta go on. you try again, and you try again.
Beyond The Sunset: How To Wring The Maximum Joy From Your Last 10 Years In Tech
- Speaker
- Vanessa Kristoff
- @vanessakristoff
- Vanessa.kristoff@gmail.com
- "Pre-tirement"
- 55 years old
-
Ageism is real
- "Will they fit the culture" BS
-
But Don't be "that person"
- "better back in the day"
-
Job satisfaction over time
- U-shaped curve
- http://www.andrewoswald.com/docs/jooparticle.pdf
-
Imposter Syndrome
- No, seriously, for reals, after all these years, you are NOT an imposter
- Hard to have fun while struggling with this
-
We are the Village Elders
- That comes with responsibility
- We should be helping people
-
It's just a job
- You should know that by now
-
Sponsor a Newbie
- Teach your coworkers what grace is, by example
-
Sponsor vs mentoring
- mentoring is "tweaking" them to fit in
- sponsoring is promoting them
-
Give whimsical talks at conferences
- Optional: create tag clouds referencing CORBA
- You have knowledge that not everybody has
-
Pretend to be an Extrovert
- Doing so will help the folks you're extroverted with
- This is why that imposter thing is important
-
Sign up for the messiest possible coding
- what do you have to fear? NOTHING
-
Don't stop learning
- Volunteer to write UI code or backend code or try that Haskell thing
- Join a new industry
-
Use your network for good
- You have contacts
- Your people need contacts
- See how that works?
-
Gossip and complaining
- It's not adding value to your work life, so just stop it
-
Make a plan
- start NOW thinking about what you'll do
- how will you fill/structure your time?
- you will need a social life once you're retired
-
Get your house in order, literally and figuratively
- Marie Kondo that crap
- Use your employer health care (if you have it) while you can
How Live Coding Changed My Life
- Speaker
- Jesse Weigel
- @JesseRWeigel
- Volunteer live coder for freeCodeCamp
-
How I got started
- Paid for private github repos rather than letting folks see my code
- "I make a lot of mistakes and I am a full time dev. Maybe it would be helpful for beginners to see that."
- First streams were really bad and nobody watched
- Asked for help on the freeCodeCamp forum
-
The community
- all ages, locations, and skill levels
- eager to learn and contribute
- overwhelmingly positive
- always learning and trying new ways to do things
- VERY diverse watchers
-
Consistency is key
- Set time for the streams
-
Keeping it positive
- Thank for negative feedback
- Ask for clarification if it's not constructive ("what can I do better?")
- People with situational anxiety, etc. feeling safe
- Video about depression, mental health issues
-
What I've learned
-
So much collaboration
- Pull requests!
-
Confidence building
- Viewers too, gaining confidence over time
- Viewers get jobs!
-
-
How you can start
- Make a youtube or twitch channel
- Broadcast and share your screen (OBS)
- Start coding!
-
Advice
- Adapt based on feedback
- show your mistakes
- show your face
- don't be afraid to say I don't know
-
It's okay to say nothing (or better yet thank you) when someone tells you something that you already knew
- it may not be new to them, they could be excited and want to help
- Always be positive and encouraging
- Give encouragement, not solutions
-
Things to avoid
- talking badly about another language, framework, library,etc
- laughing at a question (assume every question is serious)
- getting angry
- coding and reading the chat at the same time
- negativity (there is already enough negativity in the world)
Debugging Our Feelings
- Speaker
- Jamie Strachan
- @JamieStrachan
- Working at odds with professional development and depression
-
Cognitive Behavioural Therapy
- "Feeling Good" by Dr. David Burns
- parallells with developer life
- Experiences -> 👩 Thoughts -> Responses
- "There is nothing either good or bad, but thinking makes it so" (Shakesspeare)
- We can change how we feel if we change how we think
- Input -> 💻 Code -> Output
-
Starting a project, getting overwhelmed, distracted, guilty, nervous
- Similar to application in an incorrect, incomplete state. Not what we wanted yet.
- We test our output
-
We trust our feelings
- Emotional reasoning
- Because I feel something, it must be true
- Not rational
- Would be like looking at the app in that state, "I guess I'm done."
-
Feelings should be
positive- helpful
- reasonable
-
rational
- good at rationalizing, working our way back from the feelings, not super helpful
-
Mind reading
- We respond to what we think they're thinking
- Leads to impostor syndrome
- Test the feelings instead. Ask the question, don't guess.
-
All-or-nothing thinking
- Perfectionism
-
Test
-
Antiperfectionism sheet
- Task, Effectiveness, Satisfaction
-
- Testing helps identify problems, but doesn't fix them
- Code won't get any better without changes
-
Editor for thoughts
- Notebook
-
two column technique
-
left hand side: automatic thoughts
-
thought patterns that are causing us to feel that way
- "I'm overwhelmed" -> thinking, "this is going to be hard"
- "Guilty" -> thinking, I'm a procrastinator
- Nervous -> thinking, I'm not going to finish this on time
-
-
right hand side: rational responses
-
fortune telling
- we've predicted the future, and are responding as though it's true
- I can't know this
- it could also be fun, educational
-
labeling
- fancy term for name-calling
-
reduces people to one trait
- "I've always been a procrastinator, I'll always be a procrastinator"
- I'm not any one thing (I'm procrastinatING)
- I don't always procrastinate
-
fortune telling
- takes away our autonomy (it's inevitable)
- I have control over the outcome
- there are other options
-
-
- Thoughts can have a "home field advantage" in our heads
-
How does this change when it comes from someone else?
- It doesn't
- Still just an experience, just your thoughts
- "No one can make you feel inferior without your consent" - Eleanor Roosevelt
- Magnification and Minification
- Disqualifying the postitive
-
Accepting a compliment
- pause
- say thank you
-
Using the two-column technique, have someone to role play them with
- We don't always extend sympathy to ourselves
-
"Should" is the worst word in the english language
- Try "I want…"
- Test your feelings
- Get your thoughts out of your head
- Get help
Not Attended
The Mental Impact of Tech Interviews
- Slides
- https://docs.google.com/presentation/d/1y4SwAuJcYgCJ-5lSpF560QBWaMBIrlLFA5BrMH-UKFw/edit?usp=drivesdk
- Speaker
- Zack Zlotnik
Doing the next right thing
I'm really good at beating myself up.
For me, doing the next right thing means:
- Taking a little extra time in the mornings to heat up some water for tea
- Preparing myself a bubble bath and relaxing quietly for an hour listening to a favorite podcast
- Getting out for a short walk when the weather's good after a few days of feeling bad for not exercising
- Writing in my journal when I'm feeling crappy about something to get my feelings out
- Watching a favorite episode of a show when I need to feel comforted
When to sanitize and when to validate
Data should be validated coming into your application, and sanitized when going out.
For the purpose of this article, I'll work with the following definitions:
- Validation
- Checking data against the expected types and constraints of a domain, rejecting data that does not comply.
- Sanitization
- Transforming data to satisfy constraints (e.g. removing unwanted characters or escaping them) in order to allow the data through.
The part of your system that is accepting data should not need to be concerned with how it may be output elsewhere. If you can ensure that your output is always properly sanitized, you can avoid needlessly complicating your validation logic, thereby keeping your input and output layers neatly decoupled.
Reasons to sanitize output
Cross-site scripting (XSS) prevention
SQL injection prevention
I didn't want to glue my new phone mount to my dashboard
What do you do when you want to mount your phone in your car but it has trouble staying put on your dashboard? Well, if you're me, you try your hand at designing a counterweight and fabricate it with a 3D printer you're borrowing from a friend.
Automating our garage door
Now that I've got a house again, I can really start playing with home automation projects a lot more. The first thing I plan to do is come up with something to monitor and automate the garage door, as I've already panicked from not remembering whether I'd closed it when leaving the house. It turned out I had closed it, but I can definitely do without that anxiety. What better way to remove that anxiety than by being able to remotely monitor and control the door!
Choosing the hardware
Controlling a garage door is a bit more involved than automating smart lights,
so it was time to finally dip my toes into wiring up and programming some
microcontrollers. I did some research and settled on the popular ESP8266
series of microcontrollers, and found myself a set of D1 mini clones with
built-in micro USB connectors ($3 USD each). I also snagged myself a heavy-duty
looking https://en.wikipedia.org/wiki/Reed_switch][reed switch to monitor when the door is closed ($17 USD), and a pack of
3 volt DC single-channel relays ($5 USD each). I chose single-channel as I have
only one door, getting modules with more than one channel could make it easier
to hook everything up if you have more. Because this is my first electronics
project, I also grabbed myself an electronics kit with a breadboard, jumper
wires, and a bunch of fun components to fiddle around with. I tacked on some USB
cables and power bricks for powering my creations as well.
Choosing the software
There are multiple options for developing the firmware to install on the ESP8266 controller. After looking at Arduino and NodeMCU, I settled on ESPHome as its super simple to set up (Arduino coding looks fun, but I'll get everything I need just using some YAML configuration) and it integrates super easily with Home Assistant (the platform I use for all of my home automation). I was able to get up and running just by installing the ESPHome CLI tool and tossing some config together.
Wiring up a prototype
To test the module out, I wired it onto a breadboard, using it's 3.3V pin to supply power to the top positive rail and hooked its ground pin to the lower negative rail. Hooking it up to the breadboard with only jumper wires is a bit finicky (I didn't have a soldering iron at the time), so to confirm that everything was working I ran wires to hook a blue LED up between the power and ground. When everything was snug and the USB cable was plugged in, it lit up! Success!
Preparing the initial firmware
I used the ESPHome CLI wizard to generate my initial firmware configuration for
the device. The wizard prompts for various values needed for basic
functionality. In my case, I specify a name for my device (garage-door), the
microcontroller I'm using (ESP2866), the board I'm using (d1_mini), and some
WiFi credentials. Substitute in the credentials for the WiFi network your device
will connect to if you're following along.
<source src="/videos/esphome-wizard.webm" type="video/webm"> Download the <a href="/videos/esphome-wizard.webm">WEBM footage</a> of the ESPHome CLI wizard.
Review the contents of the generated YAML file, then connect the board to your computer via USB and run:
esphome run garage-door.yml
The CLI tool will generate and compile the code for you, then prompt you for the
device to install to. As this is the first time you're installing the firmware,
you must select your USB device (in my case, on my linux machine, the device was
/dev/ttyUSB0 (USB2.0-Ser!)). You'll see the logs as the device boots up and
connects to your network, and it's up and running! Not doing much yet, but it is
there and discoverable!
Adding it to Home Assistant
Now that the device is running and discoverable on the network, it can be added
to Home Assistant. This can be done on the Integrations tab of its Settings,
clicking the "+ Add Integration" button, searching for and selecting "ESPHome".
Home Assistant then prompts for the connection settings (in my case, the
hostname was garage-door.local, and the default port is 6053). It will also
prompt for its password, which is in the api: section of garage-door.yml
(the same password that was set in the wizard). As entities are added to the
ESPHome configuration and uploaded to the device, they will become available
within Home Assistant.
Wiring up the garage door detector
The first thing I hooked up was the reed switch. One wire is joined to the D1
pin on the ESP, and the other to ground. In the ESPHome configuration, I added a
binary sensor for the switch, configuring the D1 for input with its pull-up
resistor enabled, which sets the D1 state to high normally. When the magnet is
within a couple inches of the switch plate, the switch will close the circuit,
triggering a state change from high to low on the ESP pin as the current can now
flow to the ground pin.
binary_sensor:
- platform: gpio
id: garage_door_sensor
name: "Garage Door"
device_class: garage_door
pin:
number: "D1"
mode:
input: true
pullup: true
The binary sensor is using the gpio platform to read the D1 pin in input
mode with its pullup enabled. The id value will be used to reference the
sensor in other areas of the configuration, and the device_class is used to
inform Home Assistant that this device is monitoring a garage door.
I executed the esphome run command once again to load the new firmware, and
once it was up and running I was able to verify that the switch was working from
the logs as I moved the magnet up to and away from the switch plate.
I was also able to add the new sensor entity to my Home Assistant dashboard, which also updated as I moved the magnet!
Adding the garage door control switch
Next is the relay switch, which I will use to control the garage door so I can open or close it via the Home Assistant mobile app or any automations I decide to set up for it.
In the ESPHome configuration, I added a gpio switch using the D5 pin. Since
going to be activating this switch in a particular way within another control, I
marked it as internal so it can't be activated manually via Home Assistant. I
then added a cover entity representing the garage door as a whole. This wraps
up the door sensor and door control into one neat package, and lets me specify
how the switch should be toggled to activate the door.
switch:
- platform: gpio
id: garage_door_switch
name: "Garage Door Opener"
pin: "D5"
internal: true
cover:
- platform: template
name: "Garage Door"
lambda: |-
if (id(garage_door_sensor).state) {
return COVER_OPEN;
} else {
return COVER_CLOSED;
}
open_action:
- switch.turn_on: garage_door_switch
- delay: 0.1s
- switch.turn_off: garage_door_switch
close_action:
- switch.turn_on: garage_door_switch
- delay: 0.1s
- switch.turn_off: garage_door_switch
stop_action:
- switch.turn_off: garage_door_switch
Because the garage has only one switch for both opening and closing the door,
the open_action and close_action are identical. To trigger the mechanism, it
activates the switch, pauses briefly, then deactivates the switch. I used a
lambda to interrogate the door sensor's state to return whether the "cover" is
open or closed. The stop action won't really stop the door because of how the
actual mechanism works, but it is included for completeness.
The switch uses three hookups to the ESP: One from the 3.3v pin, one from the
input wired to the D5 pin, and one to ground. To give it something to control
while testing in place of the garage door opener it'll eventually connect to, I
set up a circuit with a red LED between the 3.3v power and ground lines, and
wired the relay switch in the middle. Because I only want the LED (door opener)
circuit closed when the switch is activated (i.e. a normally-open circuit), I
attached the LED circuit wires to the NO (Normally Open) and COM (Common)
leads on the far side of the switch.
Once this was done, I was able to activate the door control in Home Assistant and see the red LED toggled on and off!
By adding the cover entity to Home Assistant, I was also able to get this nifty control card!
Celebration!
Behold! The prototype works!
<source src="/videos/garage-door-prototype.webm" type="video/webm"> Download the <a href="/videos/garage-door-prototype.webm">WEBM footage</a> of the wired prototype.
Hooking it all up for real
With the prototype sorted, it was time to put it together into something I could install!
Getting it soldered
I grabbed a perforated board and got to wiring and soldering everything
together, changing the pins I was using for the sensor and controller to make
them easier to route (now D5 and D2, respectively). For my first time
soldering electronics, it went pretty smoothly. Getting the wires soldered
together took a couple of tries and aren't going to win any beauty contests, but
everything's secure and working fine.
I also added a red status LED adjacent to the blue one I used for power that will flash when booting or in an error state.
status_led:
pin: "D7"
Designing and printing an enclosure
With everything in place, I grabbed my calipers, measured everything, and hopped into FreeCAD. I put in some supports to hold the board up with some room underneath for the wires, and cut out holes in the sides where the usb power and external wires will attach to their screw mounts. I added small bumps on opposite inside walls to grab the edges of the board and hold it in place. I then made a lid that will slide in to the top, and opened holes in it for the LED to shine through and to provide access to the screws for the wire mounts. The lid also has a small bump for it to hold it in place when fully inserted.
There are pretty visible layer lines in the print, likely because I switched in a new roll of filament for this and didn't dial it in first, but where this thing is going it doesn't need to be gorgeous.
The board popped right into place, nice and cozy in its new home. The openings lined up pretty well, though I did have to widen the USB opening a bit with an X-Acto knife so the cable would fit properly.
The lid slid right in and locked in place just fine! The LEDs shine through their opening, and the screws are easy to get to and manipulate with a screwdriver.
I'm proud of my little creation and its pretty little case. Next up, installation!
Installation
Alright, time to get this thing into the garage!