I've been on the Fediverse since January 2017. I initially ran a single-user instance of Mastodon. In March 2020 I started to write Ktistec, my own implementation of an ActivityPub server in Crystal (a language with the ergonomics of Ruby but the speed of Go) because I wanted something more supportive of writing. This #introduction was written and published on Epiktistes, my Ktistec instance.

I'm an Engineer by training but now I run teams for companies in climate-tech.

I love #music, #sciencefiction and #fantasy literature (yes, I'm an R. A. Lafferty fan), attend fan conventions like #worldcon and #dragoncon, and do regular #weightlifting. I am also learning to play the #bagpipes, and I'm (re)learning #japanese.

The Ktistec executable is now ~24.7% smaller and build times are 28% faster.

I've been blogging about optimizations here, here, and here. This is the summary of the final outcome, with links to commits for the curious. I have one more post planned with a summary of my thoughts.

Here's my approach. Use nm to dump the symbols in a release build executable and then look for things that seem redundant. The first change and associated post below is a great example of what I mean—my original implementation led to the specialization of the #== method for every pairwise combination of model classes even though the result of the comparison was just false.

This might seem like a strange approach if you come from a compiled language where you mostly write all of the code yourself or invoke generics explicitly, but Crystal takes your code and does that for you. And it's not always obvious up front (to me, at least) what the final cost will be.

I've include counts of the lines added/removed because the point of this whole post is to say if you measure first and then optimize, a small change can have a big impact.

Here are the changes:

• Specialize model #==. (+7 -5)
  I talked about this here but didn't have the commit to link to. This change results in a large reduction in executable size on regular builds (~4.0%) and a small difference on release builds (~0.2%).
• Remove conversion to Hash. (+2 -2)
  This commit eliminates specialization of methods like __for_internal_use_only that get passed both named tuples and hashes by going all in with named tuples. It also eliminates instantiations of the Hash generic type itself for these cases. Reduces executable size by ~2.2%.
• Eliminate duplicate code in the executable. (+3 -3)
  This small change reduces the size of the executable by a further ~0.4% by eliminating redundant definitions of __for_internal_use_only entirely.
• Make InstanceMethods instance methods. (+1 -5)
  This was a goofy design I picked up somewhere. It's unnecessary. Changing this saves ~0.2% on release build executable size.
• Move the code for digging through JSON-LD. (+246 -281)
  It looks like a lot of lines of code changed here, but the large numbers are the result of moving code line-by-line from an included module to a utility class. Invoking these as methods on the utility class rather than as instance methods on each including class reduces the executable size by ~0.5%.
• Use map from base ActivityPub model classes. (+10 -2)
  map is a class method defined on each ActivityPub base model class. Each definition maps JSON-LD to a hash that is used to instantiate the class. Class methods defined on a base class are available on subclasses, as well. Calling the method on the subclass results in a copy of the method. This change reduces the executable size by ~5.8%.
• Move map into helper. (+104 -88)
  The map method does not depend on class/instance state. This change ensures that the mapping code is not duplicated even if a subclass's map method is accidentally again called. It looks like a lot of changes but this commit is mostly reorganization. It reduces executable size by ~0.4%.
• Replace classes with aliases. (+62 -148)
  Implementing ActivityPub's vocabulary with discrete model classes is expensive because every model class comes with machinery for type-specific CRUD operations. Enumerate aliases on each base model class (e.g. a "Service" is an "Actor"). This change reduces executable size by ~16.9%.

I'm off to optimize some queries now...

#ktistec #crystallang

After I release a new version of ktistec, I build the server commit-by-commit to see which commits increase the server executable size and build time the most. I do this because I’ve learned that small implementation details (inlined code, small methods, using blocks) can have large impacts on these numbers.

Here's the output:

Commit         Size          Time
======== ========== ======= ===== =======
248850b1   36426264          10.3
47268073   36425688  -0.00%  10.5  +1.60%
344de272   36425688  +0.00%  10.8  +3.24%
ef561f52   36425944  +0.00%  10.8  -0.08%
8ae2cbd4   36429128  +0.01%  10.8  -0.01%
3e425f3b   36429128  +0.00%  10.8  +0.22%
1487d903   36427704  -0.00%  11.0  +1.42%
935c9ceb   36427016  -0.00%  11.0  +0.14%
de37dc6a   36427016  +0.00%  10.9  -0.97%
a660a326   36427016  +0.00%  10.8  -1.12%
ff3d990e   36427016  +0.00%  10.8  +0.54%
5724a58d   36523192  +0.26%  11.0  +1.78%
7b5057d4   36523640  +0.00%  11.0  -0.44%
30ca6a3f   36541352  +0.05%  11.6  +5.73%
e2327eea   36671592  +0.36%  11.0  -5.36%
ad0d76eb   36671592  +0.00%  10.9  -0.48%
d388e74f   36671592  +0.00%  11.4  +4.59%
dacea7ad   36671592  +0.00%  11.0  -3.76%
03d5dfd8   36671592  +0.00%  10.8  -1.63%
79d9d89f   36671576  -0.00%  11.0  +1.82%
b65d292f   36792376  +0.33%  11.1  +0.95%
0ef53365   36808904  +0.04%  11.6  +4.88%
b3766e7b   36808904  +0.00%  11.1  -4.50%
56ba79ce   36825416  +0.04%  11.1  -0.50%
4824df58   36825736  +0.00%  11.1  +0.31%
c4705143   36837544  +0.03%  11.1  -0.03%
e3d37ef7   36837768  +0.00%  11.5  +3.52%
4509fa0d   36837768  +0.00%  11.0  -3.83%
0ff9237b   36837768  +0.00%  11.0  -0.55%

Overall, the server executable size increased by about 1.1% and the build time increased by about 6.8%. Maybe that's not too bad for a major feature, but let's dig in.

It's nice to see that three commits account for almost all of the increase in server executable size:

• 5724a58d Add `language` to `Object`.
  2 files +19  loc
• e2327eea Render `contentMap` on ActivityPub objects.
  2 files +17 -1 loc
• b65d292f Add translation actions to the objects controller.
  1 file +35 loc

But, compare 5724a58d to 8ae2cbd4 (Add `language` to `Account`). It added +22 loc but didn't increase the server executable size as much.

In any case, I'll look at e2327eea first. I'd like to understand why this relatively small change adds 130,240 bytes to the server executable size!

The follow ups are here, here, here, and here.

#ktistec #crystallang

i was overdue for my annual (or maybe decennial) .emacs checkup and service.

i removed support for haskell and ocaml programming—i love haskell but i'm not writing it anymore. rip. i deleted packages that support tools i've replaced with other tools (e.g. ack → rg) or i just don't use (origami, counsel/ivy/swiper, etc.).

it looks like i need to migrate from a few packages to newer packages (smex → amx, restclient → verb, etc.). smex hasn't been maintained since ~2015 so maybe it is decennial...

#emacs

listening to the repo man soundtrack... i didn't love the movie but i continue to love the tracks...!

#music

i think Zombie (The Cranberries) aged well and listening to Dolores O'Riordan sing it still gives me shivers.

"We put the exploit in a picture. The AI code reviewer never opened it."

This line from the article:

CodeRabbit ships with a default configuration that excludes image files from review outright (!**/*.png).

#ai #infosec

There are two significant new additions in release v3.8.0 of ktistec.

First, the actor cards on the followed/following pages show relevant actor status. On the followed page, it tells you how long it has been since that actor published an activity (create, announce, like, etc.) that was sent to your server. It's a proxy for how active they are (or are not). On the following page, it tells you how long it has been since you have been able to send to that server. It's a proxy for whether that server is reachable or that actor is still alive.

Second, the backend for user-defined algorithmic feeds is in place, along with a keyword/hashtag/mention feeds implementation. You can't set up a feed via the user-interface, but the feeds work if you set one up directly in the database—which is how I've been previewing them. I plan to release the frontend next week.

Here's the full changelog:

Added

• Display activity status on actor cards.
• Back-end support for user-defined algorithmic feeds.
• Apply community-relayed moderator deletes received as a Group's wrapped Announce.
• Follow a web page's rel="alternate" link when searching.

Fixed

• Avoid loading entire has_many collections when constructing child records.
• Evaluate the same-origin fetch gate against an embedded node's own identifier.
• Accept a delete of an uncached object or actor without verification.
• Catch MIME::Multipart::Error in local file-upload handling.
• Map malformed request-body parse failures to Bad Request.

#ktistec #crystallang #activitypub #fediverse