- Date: Monday 26th February 2024 at 16:00 UTC
- Location: https://meet.jit.si/Servo-TSC
- Status update
- Nico's plans
- Security policy
- Crowdfunding
- Outreach
- AOB
Attending:
- TSC members: atbrakhi, emilio, Loirooriol, mrego, mrobinson
- Other: iuriatan, matlu, nicoburns, Taym95, wusyong, zrhoffman
rego: High-level summary: A lot of work on CSS support, particularly on tables. The new layout engine has surpassed the legacy layout engine with regard to the focus folders. For the entire CSS test suite, both engines now have the same percentage.
rego: https://wpt.servo.org/
rego: Regarding other features, there was work on the Gamepad API (more to come), there is a WIP patch for ResizeObserver. There has been other activity such as improvements to WebIDL. Both stylo and webrender have moved to separate repositories to improve modularity and making it easier for others to use these parts of Servo.
rego: Regarding Android, we are still dealing with some issues with crashes in the emulator.
nico: I'm planning to work on Flexbox support. Seems like a good way to get started and Servo is missing quite a few pieces. It's not quite 100 percent sorted, but it looks like I'll have funding to work on this the next few months. Hoping to contribute some meaningful improvements.
nico: What's up with the results on wpt.fyi? The results look wrong.
rego: It hasn't been run since January due to issues with the Docker image and Servo. There are incompatibilities. We are working on a workaround for that. Hopefully we can get this running again.
rego: Hopefully new results on Sunday.
rego: They also weigh subtests and tests differently on wpt.fyi than we do our calculations, which are based on https://wpt.fyi/interop-2024
nico: Hard to tell the differences between runs.
rego: BTW the code for wpt.servo.org is at https://github.com/servo/internal-wpt-dashboard
rego: There was an issue about how to report a security issue. We discussed on the TSC Zulip channel and we decided that since there is no Servo product, these can be filed as normal GitHub issues. There was also this comment if we should set up a security policy link.
rego: servo/servo#31193
rego: https://github.com/servo/servo/security/policy
rego: Should we do this? Who would be willing to help set this up?
nico: If the policy is "create a GitHub" issue, maybe we can just put that there?
nico: Is it worth warning people that it is insecure if we haven't put too much effort into security?
rego: Like you said, maybe we can just put that simple policy there and make sure to have some GitHub labels to do that.
rego: There's a label already it seems.
rego: As we explained last time, it isn't easy to use the money from the LF crowdfunding platform. We have set up an OpenCollective page to collect these. The old crowdfunding is hidden. We are coordinating with LF to close the old page, inform people that it has moved, and transfer the money. We are also setting up GitHub sponsors to enable people to donate that way as well.
rego: https://opencollective.com/servo
rego: We are still finishing the setup for all of this.
rego: The main highlight was the talk by Rakhi as FOSDEM. The video is online (link above). Servo was also part of the LF Europe stand there.
rego: https://blogs.igalia.com/mrego/servo-at-fosdem-2024/
rego: There are a couple events upcoming. Rakhi will talk about Servo at the Open Source Summit North America summit in April and then there is the Web Engines Hackfest in June in A Coruña. There will be a bunch of people from Servo there.
rego: https://sched.co/1aBNF
rego: https://webengineshackfest.org/
rego: Regarding Outreachy we are in the process of bringing Servo back to that program. It's a very popular program and Servo used to be an Outreachy project back in the day. For this cohort, starting in May, Servo will be part of the program again.
rego: https://www.outreachy.org/
rego: There will likely be some new contributors and people playing with Servo a bit.
martin: Outreachy people suggested this time can be a bit stressful, as many people can be asking questions; maybe it's worth open us a stream in zulip just for Outreachy questions and answers, and for people to introduce themselves
rakhi: it's a great idea, as many people don't know if to create a new topic or channel on Zulip, so sdounds like a good idea
martin: I'll make it then, if people are fine
mats: great idea, maybe we can have a 1 page intro with some links for outreachy interns
rego: https://crates.io/crates/azure
rego: This crate was published years ago and the repository is archived. Microsoft approached us to see if they could take over this crate name to avoid confusion.
rego: Just checking here to make sure no one sees and issue with that.
martin: it looks any other crate depends on the old azure crate
nico: we could even publish the old crate under a different name if needed
martin: yeah, though it seems unlikely as it uses a very old version of Moz2D like 10 years old
nico: I wanted to ask about other crates that involve Mozilla. I know that Martin and Delan have been working on stylo and selectors. I am also interested in cssparser, which is managed outside already. WebRender is also in this category of being developed, but not released.
nico: What are people's plan and I'm interested in getting them released.
martin: I can give an update on our general approach
martin: our plan for stylo is to switch to the new repository, get it up to date, upstream our changes, sync again, and then think how to do releases
martin: same strategy for webrender, but focusing first in stylo
martin: cssparser has its thing already going on
emilio: cssparser it gets pulled as a regular vendor dependency into firefox
martin: do you think that's going to be the situation going forward?
emilio: cssparser is just a tokenizer, it usually get just a few changes, it's not problematic for gecko, so probably it's not going to get pulled-in soon
martin: our goal for stylo is to try to be as close as possible to upstream, to contribute changes back
emilio: I'm happy to help, I was doing the sync up to 2020
martin: some changes would be good for upstream, as we removed the difference between legacy and new layout
nico: do you think upstream would accept changes related to generics and the like?
emilio: In general that's fine, things are pretty general already.
emilio: The whole property definition thing is a bit tricky, because there is the runtime configuration of CSS properties and the the style system relies on knowing at compile time what's enabled.
emilio: Maybe this could happen using some sort of code generation or macro to work on a per-property basis.
emilio: I'd be careful, as generics can get complicated, but it could work.
emilio: everything is kind of generic in top of the DOM tree, the whole property definition thing is a bit tricky (runtime configuration of css properties), the style system relies on knowing a compile time the enabled ones, maybe a macro or something is needed, it seems tricky; happy to look at prototypes and see how it looks; seems like a good direction in general if it doesn't harm maintainability; I'd be careful but seems like a reasonable direction
nico: that makes sense
nico: What the situation with WebRender?
emilio: My understanding is that it's like Gecko.
emilio: I don't think anyone is publishing that or making it useful.
martin: the plan is basically the same than for stylo, but we have to get up to date in order to be able to release a new version
martin: we've coordinated with the graphics folks at Mozilla, things will be versioned in the downstream repository; which requires us thinking with upstream
nico: probably there could be documentation and examples for these creates, for releases
martin: that would be super useful yeah
mats: I noticed there was an announcement about Tauri 2.0. I saw there are plans to do mobile including Android.
mats: I wonder if there are any plans to replicate those demos on Android?
rego: For Servo, we are still having issues, but when those are solved maybe trying those demos on Android would be good. I'm not sure if Yu Wei has any plans or not.
yu wei: I haven't tested it yet.
yu wei: For the beta, I only did the early (?) part, which is basically using Chromium / Android Web View.
mats: Maybe we can have an AP for having some ideas in the next TSC meeting around Android?
yu wei: One difficult part of mobile development is that there aren't many good rust development tools.
yu wei: Maybe at some point we could work on making the Rust mobile tooling ecosystem a bit more mature.
mats: I agree that could be good. We have some ideas for improving tooling in Rust.
yu wei: Feature request in mozjs. I want to push the prebuilt version in SpiderMonkey. Is there someone I can work with to push this PR?
martin: Let's arrange a time with Mukilan to walk through this if others aren't available to review it.