And second, a lightning talk about the work I have been involved to improve the V8 @v8js
support for native stack walking profiling.
See you at BlinkOn!🧵(3/3)
First, a breakout session about Linux @ChromiumDev running in low memory devices. Can Chromium run in a 512MB device? How can Chromium better support those scenarios? 🧵(2/3)
New blog post: "Maintaining Chromium downstream: upgrade strategies"
https://t.co/fPvXNKqU48
🧵(1/3) Continuing the series, this post dives into a crucial aspect: keeping track of the upstream #Chromium repositories changes. cc @igalia@ChromiumDev
🧵(2/3) Which branch should you track? How often? #Git Rebase vs. Merge? All these questions will sound familiar if you are maintaining a downstream of #Chromium. The post reviews the problems, and gives some tips. Enjoy!
These are not small contributions. And this is not an "overnight success". It is the result of years of hard work on the domain including major contributions to the underlying @v8js engine.
I'm proud that @techAtBloomberg and @igalia get to support @JoyeeCheung's work on Node.
My colleague Jani has published a blog post on the work he's been doing lately at @igalia to get @WPEWebKit working on Android, thanks to the support by @NLnetFDN. It summarizes the latest changes & next steps, and includes some demos too. Check it out at https://t.co/qznw3CmxMH
New blog post! Maintaining downstream @ChromiumDev https://t.co/F3TSjgCPtW
Why is Chromium used in different projects? Why do they need a downstream?
This is the first post of a series about the challenges of the projects that maintain a downstream of Chromium.
cc @igalia
@leszekswirski@igalia@TechAtBloomberg That's interesting! My observations with launch time, without recording ETW or enabling Linux Perf ,was that --interpreted-frames-native-stack had no impact in launch time. But those were quite informal tests. What could I use for exercising launch time metrics (as LCP)?
The blog post runs benchmarks to know how much overhead instrumentation has, and if it is possible to enable those by default (short answer: not for Linux, maybe for Windows). This work is part of the collaboration between @igalia and @TechAtBloomberg. (4/4)
Unfortunately, in @v8js, profiling instrumentation is disabled by default. Why? Profiling instrumentation requires sharing the code addresses (with extra files in Linux Perf, or specific tracing messages in Windows ETW). That has an impact in memory and CPU usage. (3/4)
@webos_ose 2.24.0 is out!
Main highlights from web stack side:
It is now based on @ChromiumDev m108. And we now use LLVM Clang and libc++, instead of libstdc++ and GCC. That way we align better with upstream, far more tested with this toolchain.
https://t.co/txTzkeYlEG