I’m trying this format, after reading with interest the notes from a few friends in different circles. Let’s see if I can stick with it.
This week, I was back from an extended period off of work, to finish the move to my new apartment in Paris. The walls are now without any holes, extremely clean matte white, the power well distributed in the office, CAT6 cable is running from the router to the office and a nice access point covers the whole apartment with ease. It was clearly time to plug the Mozilla workstation to my new fancy power outlets and get going.
After fixing a few test failures I had left intermittently failing (with an acceptable rate), and that were seemingly puzzling, it took a good 5 minutes each to come up with the patches for bug 1686295 and bug 1693070. Less pain for the build sheriffs, and a good reminder that sometimes you’re just too tired to fix very simple bugs. I’ve written a few other Gecko patches this week, related to disagreements between a codec and its container, crash fixes in weird audio device switch corner cases, and made lots of comments on lots of shared documents, as usual.
I also updated a library we use for time-stretching called SoundTouch in bug 1677881. In addition to fixing an assert failure, this gets us faster code when compiled for ARM64 because the maintainer has written C++ code that is well auto-vectorized by modern compilers. This is significant: in addition to mobile, ARM is starting to get traction on desktop computers now.
Most of my time this week has been spend on web specification-related matters. 7.5 years after I became an editor of the Web Audio API specification, the group has finally fixed all issues reported to the text, and the Web Audio API is going to become a W3C recommendation. It’s almost 200 pages of very technical stuff when typeset with a small font on A4 pages, to put things in perspective.
In practice, we’ll all continue adding more feature
to the specification, and this already happens on the V2
repository, come join
shaping the future of audio on the web it’s a lot of
In other spec news, we’re full-steam ahead on Web Codecs, and I’m trying to focus on two aspects I find particularly important:
- Auditing the API in terms of memory-related performance issues, where allocating memory is forced by a particular API pattern, or authors are required to copy memory, or API are required to internally copy memory
Apart from that, the people working on Web Codecs are trying to make sure the API will cover all use-cases, so we’re meeting with people from various domains that use audio and video codecs, and they tell us what they need and we take notes, roughly.
I’ve recorded a rather bassy mix that goes from Tokyo to Durban, and laces mostly techno with lots of textures and Gqom, with hints of UK-hardcore or UK-hardcore influenced bits. It’s probably not for everybody and is better enjoyed on a system of a decent size or headphones with a good bass extension.
It’s called La reprise du travail and is about an hour and a half of kick drums. There are a few annoying mistakes, but it wasn’t really prepared or anything and I’m not exactly experienced.
I think I started Ableton Live once or twice, but nothing came out of it. Maybe two photos were taken. Energy levels are still at an all-time low. We’re back in a pseudo-hard-curfew-lockdown hybrid thing, it’s annoying but it is how it is.
Cooking-wise, a very nice dish of Lasagna al ragù Bolognese was made, this time with dried pasta and not homemade spinach pasta. It’s better, in line with my current thinking that dried pasta are most of the time (but not always) a superior choice. The leftover ragù disappeared promptly with rigatoni. I’m starting to experiment with brick pastry sheets and made the classic ricotta-spinach-egg-onion-pepper-salt filling, it was decent but has potential. My real quiche Lorraine game on the other hand is starting to be really solid.