Disqus - Latest Comments for zellyn

Re: I keep making things out of checkboxes | Bryan Braun - Designer/Developer

zellyn — Fri, 01 Mar 2024 11:46:09 -0000

Check out iq's videos and tutorials on signed distance functions… you'll get nerdsniped again, and they'd definitely work for checkboxes. Think that video with the moving cubes made of cubes, but easily rendered from scratch!

Re: You Should Be Wearing Sunscreen on Your Face Every Day

zellyn — Fri, 11 May 2018 10:22:28 -0000

Do you have a recommendation that doesn't kill coral?

Re: Cthulhu: Organizing Go Code in a Scalable Repo

zellyn — Wed, 11 Oct 2017 11:47:08 -0000

Your numbers are similar to ours at Square. Not counting `vendor/` and `protos/`, I get:
```
Language files blank comment code
--------------------------------------------------------------------------------
Go 4237 130782 61022 796953
```

You might be interested in https://github.com/square/g... if you use protos: it tries to make protoc act for Go generation the same way it does for other languages.

We've actually been recently prototyping a `gta`-type analysis for our builds. We've lasted this long without it only because we shard our builds pretty heavily, using https://github.com/square/k..., but testing only necessary packages promises huge speedups.

Re: Monorepos: Microservices in Disguise

zellyn — Tue, 18 Jul 2017 10:02:49 -0000

While it's certainly possible (and probably useful in the short-term) to take advantage of the immediate benefits of the naive approach you describe, this is not the way Google (one of the main monorepo examples) nor even much smaller companies (like Square, where I work) use our repositories. (Actually, at Square, we have a Java monorepo, a Go monorepo, and lots of Ruby repos.)

But our services (despite being in the same repo) have completely separate lifecycles, API versioning, etc. Deploying everything at once when you have hundreds of apps would be impossible.

The big benefits of a monorepo are having atomic changes across all code, and being able to trace out and test all code that depends on a change. This is particularly useful when modifying API definitions (protocol buffers, in our case). For instance, when you modify a proto that lives in the Java repo, it's very easy to re-run all tests that depend on that proto. However, you'll only find out later if you broke Go or ruby code with the change. If we had everything in one repo, the CI tests would tell you immediately.

As far as waiting for master - there's indeed some of that. But our CI system is smart enough (for Java at least) to run only affected tests, so the set of tests run tends to correspond to the app you're working on, unless you're changing core code.

Re: Go Advent Day 17 - Pond: a New RSS+Atom Syncing Protocol

zellyn — Tue, 17 Dec 2013 12:05:14 -0000

You know about https://github.com/agl/pond, right? I think having the same name is going to cause confusion...

Re: OS-9 Keyboard Codes: Brandon Rhodes

zellyn — Mon, 25 Feb 2013 12:54:30 -0000

I guess we're both the right age to have old-computer nostalgia. :-)

I've recently been spending a lot of time reading old manuals, etc. - except my particular flavor of nostalgia is the Apple ][+. :-)

Results of my nostalgia kick are avaliable in very preliminary form at https://github.com/zellyn/g... and https://github.com/zellyn/g... (nothing interesting in the second one just yet).

Re: Imagine Never Hearing the Phrase 'Aliasable, Mutable' Again

zellyn — Wed, 21 Nov 2012 00:48:03 -0000

The one place that does not seem simpler to me is &mut, which would now mean “the only mutable reference to the data in question” rather than ”a mutable reference to the data in question”.

(Speaking as a distant Rust observer), that doesn't seem too complicated. Unless I'm missing something, it sounds like, "Immutable by default, mutable only if you hold the (single) magic ticket." :-)

Re: Craig Venter Wants to Solve the World’s Energy Crisis

zellyn — Fri, 18 May 2012 17:42:28 -0000

I caught part of this on the radio. Is there a link to a recording?

Re: UITableView Basics - NSScreencast

zellyn — Fri, 17 Feb 2012 15:36:37 -0000

I think you left reuseIdentifier nil even after modifying to use a reuse queue.

Re: f.lux: F.A.Q.

zellyn — Fri, 10 Feb 2012 21:30:12 -0000

When f.lux has been running a while, it seems double-red. Quitting and restarting it fixes things. Happy to post any debug info you need.

Re: Graphene improves lithium-ion battery capacity and recharge rate by 10x

zellyn — Wed, 16 Nov 2011 12:37:26 -0000

Is 50% dropoff in 150 charges comparable with current battery technology?

Re: /2010/jan/08/new-home/

zellyn — Fri, 08 Jan 2010 16:07:15 -0000

Nice! Do you have an rss/atom feed?

Re: I want your awesome python snippets.

zellyn — Sat, 19 Dec 2009 12:16:22 -0000

Perhaps too long, Norvig's sudoku solver is beautiful: http://norvig.com/sudoku.html

Re: The Kember Identity : Elliott Kember dot Com

zellyn — Tue, 19 May 2009 08:38:22 -0000

For fun, you could keep a tally of the shortest “Kember Loop” found so far… assuming anyone has found one.

Re: Ruby-style Blocks in Python

zellyn — Mon, 09 Mar 2009 10:03:35 -0000

Ummm. I think you need better examples of compelling uses of blocks. I'm not sure exactly how select() works, but it sure looks like you're using a boolean filter-function for its side effects. Why not just use "for employee in employees.select(): …"?

Re: How FriendFeed uses MySQL to store schema-less data - Bret Taylor's blog

zellyn — Fri, 27 Feb 2009 09:41:17 -0000

Just curious — did you consider any of the BigTable equivalents?