cmr's devlog

Not bad for a morning's work. Yay python!

I've written and released hprof, a small hierarchical profiling library in Rust. Not sophisticated, but useful.

Today I played with Ember. It's quite a neat framework, and handles all of my needs for Starclient. I've gotten the hang of general workflows using it, through implementing the classic TodoMVC application. I also read a good chunk of its documentation. It's very well engineered, and seems to capture all the behavior I would have written myself. The documentation is also fantastic. Tomorrow I'll continue experimenting, and start thinking about the UX.

So, for whatever reason I guess I forgot to update this. From the 19th to the 26th, I indeed managed to do everything on my Rust todo list for this break, as scheduled. On the weekend, I worked through TAOCP 1.1 and 1.2.1-2. I also read two of the assigned books for Cultural Anthropology, which I am taking next semester. My goal for the next week and a half until the next semester starts is to at least read all of the material that will be covered, if not do some exercises. I've found that when I try to rush through the semester, I don't have quite enough time to think and ponder and tinker to produce results that satisfy me. Hopefully by getting the gears churning early, I can get closer to the much-envied 4.0 GPA (I managed a 3.2 last semester, which is a drastic change from the 1.5 of the semester prior to that).

Everything is completely setup and working, yay! I started doing a bunch of things today, but kept getting sidetracked and distracted. Not a very productive day. I have, however, almost finished https://github.com/papers-we-love/papers-we-love/blob/master/plt/fundamental-concepts-in-programming-languages.pdf?raw=true and I don't plan on sleeping until it is done.

RBAC policies all setup! grsecurity's learning mode is quite excellent, and the policies themselves are in a very understandable format. I'm now running firefox outside of a VM, because running it inside a VM had mediocre performance (which I am willing to put up with) and miserable CPU usage/battery life even when idle, which I am not willing to put up with. I set its policy so that it can't touch any files outside of ~/Downloads and its profile/cache directories (~/.mozilla). I'm still worried about graphics-stack related attacks, so I've set webgl.disabled to true, which closes up web-accessible GL. I refuse to install Java on the root OS, and proprietary software such as Flash is obviously out. Chromium has more strict sandboxing etc than Firefox, but Firefox is still improving on that front, and I prefer it. Maybe soon I can use Servo as my day-to-day browser...

Almost finished setting up the laptop. I'll write a blog post about it and the attacks it hopes to mitigate. It's a pretty cool setup I think. And since basically everything I'll be doing is in a VM, I gain a TON of flexibility in terms of save/restore/snapshots, basically for free, with great dedup.

Laptop came, fought with its UEFI firmware to try and send a damn ATA SECURE ERASE to it. Gave up, sent email to Asus, and not I haven’t done anything productive at all today! :grumble:

Finished up fallback with Niko's help, I made some dumb, obvious mistakes. Renamed imem/umem to isize/usize to see how it'd look. Also wrote a "fairly generic" cycle collector, https://github.com/cmr/rust-cc/blob/master/src/lib.rs. I couldn't figure out how to avoid the box, unfortunately.

Anxiously awaiting a smaller laptop that's arriving tomorrow, the almost-16" thinkpad doesn't suit me very well. I ordered a zenbook. ThinkPad W540 looking for a good home, asking price $2000 or best offer!

Moved to uni, classes start tomorrow. No visible work, but I'm about 90% on the planet representation, starting prototype tomorrow.

Lots of thinking about how to represent an Earth-scale planet efficiently, esp coordinate + distance computation wise. I think I have something, more when I have the math nailed down... Good conversation with bjz about generative art/world generation!

Been a long time since I've updated my devlog! I plan on doing it a lot more consistently. Today I spent a lot of time experimenting with CEF, wanting to use it for handling the GUI in my Super Secret Game Project (hereby known as "SSGP"). I decided it's a very large dependency that I don't care to have just yet. So, I spent the remainder of my time thinking about the best way to implement a GUI library, including evaluating existing toolkits and approaches. I think the "immediate-mode GUI" approach that is currently in vogue on HN etc has lots of problems that, when solved, devolve into requiring a more complex implementation than any "retained-mode GUI". Maybe the interface simplifications are worth it. I'm not sure, but I'm going to give a shot in the context of conrod, which I'll have a PR for tomorrow.

A nice, relaxing weekend. Struct undefined-representation patch almost done, just fixing fallout now. Submitted an RFC for some changes to the lexical syntax. I’ve been reading a bunch of resources about formal languages, and bought a copy of the Dragon book. A friend and I wrote optimizing brainfuck interpreters in parallel. Mine is of course in Rust, and is 33% faster than his C++14 one. His does more optimizations and has roughly the same structure, so something rustc is doing gives it an advantage.

Still deep in the lexer, about 70% done my changed. Also fixed up the enum size lint and submitted my new languaged design faq.

Waist-deep in the lexer. Refactored everything to use methods instead of functions taking a &mut StringReader. Redefining the token types to store strings instead of numbers (will re-introduce that in the parser).

Today I spent working on the enum size lint, updating the test-shard fixes, and reviewing the lexer and parser. Overall, not an entirely productive day.