Apr 25, 2025 | RUMCG Meeting

Attendees

Nic Jansma, karlijn@rumvision.com, Cliff Crocker, Andrew Creskey, Barry Pollard, Brian Kardell, Edwin Molina, Eric Goldstein, Erwin Hoffman, Henri Helvetica, Issac Gerges, Ivailo Hristov, Jose Dapena Paz, Joseph Scott, Magnus Dahl, Michal Mocny, Philip Tellis, Robin Marx, Simeon Totez, Stephen Pierzchala

Agenda

RUMCG Issue Tracker
Funding work for browser features

Admin

Next meeting: May 9 2025

Minutes

RUMCG Issue Tracker

Nic: We've started to populate a Github Project with top issues the WG is tracking
... https://github.com/orgs/w3c-cg/projects/1
... Hope to use that to try to prioritize what we want to focus on

Funding work for browser features

Cliff: Alluded to this in the earlier discussion
... Kicked off from WebPerfDays Amsterdam
... Premise from why we put this group together
... Wanted to have a follow-up session from that, getting stuff done
... Specifically out of that came a few approaches
... How do we get moving?
... Shame UAs into doing it?
... Look at other ways of creating use-cases? Input for why this work needs to get done?
... Didn't talk about Interop, but that became huge
... Safari folks may have championed that
... Real one that stuck out, hair-brained idea, we have a lot of us working for different-size companies
... Could we look at funding some of this work?
... Way to crowd-source funds to be able to go out and hire someone, or consult, to build features we care about/
... Interesting discussion
... Wanted to bring that here to the group
... Is that something that has legs? Could work in other cases
... Have Igalia here, group that has done this, and features we care about in browser, for hire
... What do people think about this idea?
... What would it look like at your organization? Challenges or difficulties? Be able to champion?
Henri: Didn't Yoav do this a few years ago?
... Around responsive images?
Brian: Yoav did try to crowd-fund years ago, for Responsive Images CG
... Was also in process of switching jobs and going to do it anyway
... He didn't really raise enough money to do it full, a few thousands of $
... Other attempts to crowd-fund some things
... A lot has come since then though
... Back then, no Github sponsors
... Open Collective
... Part of this does come from us talking
... [Igalia Introductions from Brian and Jose Dapena Paz]
... Experience in Container Timing in Chromium, some other colleagues couldn't make it working on WebKit/etc
... We work on web platform, number two contributors to WebKit (by a longshot, 16% of all commits come from Igalia)
... #2 in Mozilla
... #1 in Servo
... Until recently #2 in Chromium (until MSFT passed us)
... We do a lot this, through contracting companies, with needs being met
... Priority question is also relevant in terms of spec'ing and attention things get
... Roughly thinking the longer it is between spec'ing and dev, etc -- not enough scrutiny creates problems
Nic: Funding is a challenge, but could be worth it. Could be political barriers to bringing a funding request in various companies. But we should have a conversation and at least attempt it.
Eric: What it sounds like is money is an issue, Apple certainly does not have money issues, not saying we should say how they should spend their money
... How do engineers respond?
... Where is pushback?
Brian: There are a finite number of people on all of the teams
... If I show up and really want a thing that Chromium isn't interested in, because they're already using their people
... How you direct these things is tricky
... Historically Chrome has had a 10x investment
... They do a lot of R&D of web platform
... They invest money and don't necessarily want to make changes
... Challenges, things that don't come up in those conversations
... It's scrutiny and scrutinizing those things
... Some things related to privacy, security, not being caught because there's not enough eyeballs
... Time and battling for attention, which is precious
... Web engines underscore everything
... Every screen you touch is running a web engine
... They have a lot of needs, use cases are everything
Nic: Don't want to make this just a Safari issue though
... While we focused on INP / LCP in Safari, that's just one example
... For example we could pitch idea and experiment in one browser
... Igalia/ContainerTiming with Bloomberg is a good example
Jose: Bloomberg Container Timing isn't only case, they've been investing a lot in Web Standards
... Identify issues, propose solutions, CSS, etc
... Working on telemetry integration improvement in Chromium
... There's an idea, trying things, trying to find out if there's more interest, process to get into working standards
... Jase Williams from Bloomberg, myself, doing experimentation and polyfills
... We tried an implementation to see how complex it would be
... Landed in Chromium, implementation was simpler than I would expect
... Important to see and try what can be done
... Chromium has Intents to Experiment, Origin Trials, etc
... In other browsers, not sure as much how it works
... Experimenting and contributing back to web community, we could find something useful for more people
Cliff: I have some questions around how folks work with Igalia
... One thing I'm trying to get my head around, how does this work?
... Let's say I want to see Container Timing in WebKit
... Awesome in Chromium
... Want to see it in WebKit
... Who do I call, ask Brian for a quote?
... Sounds like a lot of discovery, experimentation, unknown
Brian: We work in time and materials
... Nobody wants to go into something informationless
... We do some research, idea on estimates and t-shirt sizing
... Due-diligence on difficulty on other implementations
... Talk to people have have experience
... May have some preliminary discussions with folks at e.g. Apple, see if they're opposed
... A lot of this is talking about priorities, come to the table
... If a browser doesn't have resources allocated
... Sometimes if one vendor implements, it may excite the other vendors
... Intent to Ship is a forcing function for others to look at
... A direct conversation with some people to see if they're opposed, wanting to work
... Sometimes there's a standards-position that may cause a "no"
... If we're talking about something as a group, that's not the model
... Igalia has a MathML Collective
... Also for Servo
... Have to get a lot of small contributors
... Depends on where we think the size is
... Come up with a goal and try to raise towards that goal
... Some sort of sustaining funds, where you can direct how you want
... Priorities is important for this group too
... Say you have $60-90k, have to decide how to fund it
Cliff: One of the things I was hoping coming out of this discussion, what's the path forward?
... Do we work with someone like you directly Brian?
... The mechanics are important here so we can get our heads around it
Brian: Servo Collective
... Depends on how much money you raise, maybe it's just enough for bug fixing
... The main thing is decide these are the voting members, and whatever you decide how to direct $
... Basically it's just come talk to Igalia and say here's the $ you have, are you willing to work on this
Cliff: For this group, we tried to do this before with WPO for WebPagetest
... We did get some sponsorship for
... I'll take an action to review with Pat^[a]
Nic: For existing collectives, is it any big companies or just all individual contributors?
Brian: For MathML and Servo, it's all small
... We have some other funding besides that
... Igalia has an Open Prioritization experiment
... https://bkardell.com/blog/OpenPrioritization.html
... 80% of actual funds were through a few larger donations of $5k or $10k or $15k
... 150-200 donors, but most came from larger contributions
Nic: That's good to know and I think we represent in this group many companies of different sizes. But I don't think anybody here is expecting any particular company to want to pony up a very large amount of money for something like this. But if we could do a few five thousand, ten thousand dollar commitments, that may be feasible. Worth exploring a bit.
Cliff: Worth looking at back-log and features
... We're in a holding pattern with LCP and EventTiming
... Maybe we can look and grab something
... Maybe we could look at one or two from backlog and get a t-shirt size for something like this
... Don't want to just get $Nk and not be able to scratch the surface of something
Brian: Would have to do some work with colleagues
Cliff: Maybe some takeaway for this, what has legs?
... Do people care enough about a specific features
Brian: Container Timing specifically?
Cliff: Selfishly yes, but I think it's a good example
... Are there other things?
Nic: The rallying point for us back in the fall was these two things that have been picked up as part of Interop 2025.
Brian: Straw poll on interest?
[about 10 thumbs-up from various companies]
Ivalio: Do we know what will happen in Interop?
Brian: I'm on Interop committee, we choose things that are ambitious but achievable goal
... Would expect effectively, we're all intending to work on that. Not a promise, but can expect a big jump in passing tests in Interop in all of the browsers
Cliff: No formal statement or commitment, e.g. from Apple to have something done by a specific time?
Brian: Proposals haven't usually failed
Ivailo: COmmitment for LCP?
Brian: For Interop, we get submissions from the community. We look together to see what we cna get in Interop. Look at tests, scoring mechanisms.
... Things in there are the priorities.
Cliff: Posts some wrap-ups on Interop 2025
... https://web.dev/blog/interop-2025
... https://webkit.org/blog/16458/announcing-interop-2025/
Barry: Browsers all have Veto authority, if we get past that, likely but no guarantee
... We're thinking INP/LCP delivered to browsers by Dec 31
... Commitment they give by submitting to Interop
... No guarantees
Nic: Can some of the funding go towards spec work in Working Groups? I think that's one of the reasons why Container Timing has been moving forward so well, is constant discussion of its progress in WebPerfWG
Brian: I think it's often like that, priority is often in spec'ing
... When you can get multiple people in multiple engines interest and focused at the same time, the WG gets more productive
... Don't want to get things wrong, have WebCompat issues
... Will find holes in spec, questions, where it says X but implementation is something else
.. You have to take to WG
Cliff: Thanks to Igalia for joining
Brian: Focus-Visible in webkit was final implementation
... https://blogs.igalia.com/mrego/tags/focus-visible/
... Need to add new WebPlatformTests, spec questions, etc
... Was also crowd funded

AI notes

Overview

The RUM Community Group (RUMCG) meeting focused on funding browser features and improving the group's operational framework. After introductions and administrative updates, Nic Jansma introduced a new GitHub project for tracking issues related to user agents and feature requests, aimed at better prioritizing future discussions. Cliff proposed a crowdsourcing model to finance developer initiatives, a concept sparked during the Web Perf Days in Amsterdam, with insights provided by Brian Kardell and the Igalia team on their funding strategies and collaborations with browser vendors. The team explored practical considerations for this funding model, including the importance of selecting specific features for potential contributions from larger companies. Discussions also highlighted the interplay between funding and ongoing interoperability initiatives, with a commitment to follow up on action items and an agenda set for the next meeting on May 9, 2024, to delve into HTTP header adoption. Various members were tasked with specific follow-ups to advance the group's efforts in development funding and feature prioritization.

Notes

️ Introductions and Setup (00:10 - 05:10)

Meeting was recorded with auto-record feature enabled
Nic Jansma confirmed recording was active for all participants
Meeting slides were shared with agenda items
Next meeting scheduled for May 9th, 2024 at 10am
Next meeting topic proposed: HTTP header adoption (time, origin, server timing)
The group maintains a backlog of topic ideas that members can vote on

Issue Tracking Project (05:10 - 15:02)

Nic Jansma showcased the new GitHub tracking project 'Rum CG Tracking Issues'
The project is organized with categories for specific user agent issues, feature requests
Currently only admins (Cliff, Caroline, Nic) can add issues but they're considering expanding access
The issue tracker can help prioritize discussion topics for future RUM CG meetings
Project will serve as a bridge between RUM CG ideas and the Web Performance Working Group
Eric Goldstein expressed interest in contributing ideas to the tracker
Cliff suggested potentially expanding the GitHub project to include meeting agendas and minutes

Funding Browser Features Discussion (15:02 - 26:42)

Cliff introduced the idea of crowdsourcing funds to hire developers to build features the group cares about
This concept originated from discussions at Web Perf Days in Amsterdam in November
Brian Kardell and team from Igalia joined the call as experts in this area
Brian explained that Igalia is currently working on container timing funded by Bloomberg
Brian mentioned they're the #2 contributor to WebKit (16% of all commits)
Nic noted that budgets are tight everywhere, making this type of funding challenging
Eric questioned whether the issue is that browser vendors (especially Apple) are unwilling to prioritize certain features

️ How Igalia Works with Browser Development (26:42 - 42:40)

José Dapena Paz from Igalia explained his work on container timing for Bloomberg
José described how they first experiment and create proofs of concept before implementation
The process typically involves identifying interested parties and allocating resources
Brian explained that Igalia works on time and materials basis with preliminary research
They provide 'T-shirt sizing' estimates after researching implementation complexity
Igalia typically has preliminary discussions with browser teams to gauge opposition
For group funding, Igalia has collectives (MathML, Servo) but most significant funding comes from larger contributions ($5K-15K)
External funding works as a 'forcing function' that makes browser teams reassess priorities

Practical Considerations for Funding (42:40 - 50:34)

Nic asked whether any of Igalia's collectives had received major company contributions
Brian noted larger companies typically contract directly rather than through collectives to avoid fees
Cliff suggested selecting a specific feature (like container timing) to get a cost estimate
The group discussed identifying specific features worth funding before approaching companies for contributions
Brian suggested doing a straw poll to gauge interest in the funding approach
Ivailo asked about the relationship with Interop 2024 initiatives to avoid duplication

The Interop 2024 initiatives aim to streamline web performance metrics across different browsers to ensure consistency.
There is a focus on aligning priorities among browser vendors to prevent redundant efforts and enhance collaboration on shared goals.
Regular updates and feedback loops are established to monitor progress and address any discrepancies in implementation across browsers.

Interop Initiatives and Next Steps (50:36 - 58:55)

Brian (part of Interop committee) explained that Interop features have high likelihood of completion
Barry noted browsers have veto authority but once accepted, there's full intention to implement
LCP and INP metrics are expected to be delivered across all three browsers by December 31st
Nic asked about funding for specification work, not just implementation
Brian confirmed implementation work almost always identifies spec issues requiring working group involvement
Next meeting confirmed for two weeks later (May 9) to discuss HTTP headers
Meeting closed with an action for Cliff to follow up with Pat about previous WPO funding model

Action items

Nic Jansma

Add interested members' issues to the GitHub tracking project (14:38)
Post recording to the RUMCG agenda Google Doc (03:26)
Contact Web Performance Working Group about issues of interest after their triage (14:21)

Cliff Crocker

Follow up with Pat about previous WPO funding model (42:51)
Request a T-shirt size estimate from Igalia for Container Timing implementation (46:12)

Brian Kardell and Igalia team

Provide cost estimates for implementing Container Timing in WebKit after consulting with WebKit colleagues (46:29)

Eric Goldstein

Reach out on Slack with ideas for the tracking board (14:43)

[a]@cliff@speedcurve.com to review with Pat

_Assigned to cliff@speedcurve.com_