@azonenberg when I was trying to do some hacking on ngscopeclient I wondered about the slow compile cycles. It looked to me like the ngscopeclient.h header was pretty huge (and included from a lot of places).
Wanted to do some testing to see what the impact was, but `cmake -DUNITY_BUILD=ON` was the easier fix for me and I never investigated...
The #Rust support in the #Linux #kernel is now officially a first class citizen and not considered experimental any more:
https://git.kernel.org/torvalds/c/9fa7153c31a3e5fe578b83d23bc9f185fde115da; for more details, see also: https://lwn.net/Articles/1050174/
This is one of the highlights from the main #RustLang for #LinuxKernel 7.0 that was merged a few hours ago ; for others, see https://git.kernel.org/torvalds/c/a9aabb3b839aba094ed80861054993785c61462c
Digitale Unabhängigkeit in deiner Hand!
Am 01.02.2026 ab 15:00 Uhr laden wir euch zum C3PB e.V., Westernmauer 12-16, Paderborn ein.
Thema: Sichere Messenger – wir schauen uns gemeinsam Signal als WhatsApp-Alternative an.
Der Digital Independence Day sensibilisiert für Datenschutz und unabhängige Alternativen zu Big Tech.
Die Veranstaltung ist kostenlos, Essen & Trinken vorhanden, Spenden willkommen :)
Weitere Infos auf: https://c3pb.de/category/blog.html#Digital%20Independence%C2%A0Day
@rburchell since you claim fragmentation and arena hits you seem to be dealing with a churn of fairly large objects. So any tweaking of the knobs will likely just be tweaks on slow-ish allocator paths.
IMHO fixing/improving the allocation behavior is typically the better return on investment.
@rburchell tough to reason about things without experimenting myself... So mostly asking the somewhat obvious questions:
1) What is the metric that is tanking? Whole system eventually running out of memory?
2) Can you reduce the allocation count/churn?
3) Can you fix the short running threads with a thread pool of some kind?
Thorsten Leemhuis (acct. 1/4)
uwu1ba
C3PB e.V.