æʃliŋ, medical grab-bag @aescling@cat.family

what is it about UCSPI servers that seems to make proxying to them break client expectations?

why does curl get near instant responses from domains served by my httpd but not web browsers and various other clients? why does firefox repurrt goddamn seconds of latency?

@raindrops i remember somehow just knowing they were doing to do something terrible eventually but i still hate to see it

uv is great dammit

https://astral.sh/blog/openai ugh

mornyawn btw . also @clock, what time is it

in software, “lightweight” is a relative term

forgejo uses as much memory while idling as my xmpp server does while actively connecting to other servers lol

@aschmitz to be clear, though, my issue is with the implementation, not necessarily that checkout is an optional action. i don’t get why you need to run typescript just to git clone, is my point

@aschmitz i suppose, yeah

gitlab CI feels a lot more reasonable to me

the actions/checkout action has me rather convinced that github actions is seriously overly convoluted bullshit. isn’t it a bit overkill to run a whole-ass typescript purrogram just to make the current repository available fur your current github action to work on?

my nginx config fur gitweb and my gir-daemon config do not clone my forgejo repos purropurrly but forgejo/nginx handles https cloning anyway so it’s not the worst

i’d like to unbreak the config though

ny’all, including you, @clock

@wallhackio bumby like ball go fast

@coriander oh. that. yeah