Show newer

looking at a flamegraph in despair, realizing there's nothing major I can actually change, off the top of my head, but also realizing the cost of certain stuff.

33% of application time is spent _just_ on serde_json::de.

10% on alloc::sync::Arc<T>::drop_slow for a single type of thing.

And a whopping 42.47% on tokio::runtime::task::raw::pol<tokio::runtime::blocking::task::BlockingTask>, which makes me wonder if I'm just misunderstanding something, or whether Tokio just has a *ton* of overhead.

I don't really know where to go from here. I can't really make anything faster than serde, and trying to shove those serde_json::de into the blocking task pool didn't change anything, so I assume it's already being done async enough.

I don't really know what the deal is with that Arc::drop_slow, but that's only 10% and I'm hoping to get *at least* a nice solid 30% improvement before I feel like I'm doing something worthwhile.

Anyone know Tokio enough to understand a flamegraph for something that uses it?

#rust #rustlang

"believe victims" 

Seeing people talking about "believe victims don't go looking for evidence before blocking".

This is an easily weaponised policy unfortunately and it is frequently unclear who is a victim.

Believing victims is important. When someone says they've been hurt, believe them.

Buy when implementing harm reduction, the rule should be:

TRUST, BUT VERIFY.

So I was checking my logs and people report me to me all the time, but I got a report from mas.to, who has defederated with me, which is fine, but what was said got my attention.

They called me racist and said I didn't use content warnings for my content, which we've discussed at length this week, and how entitled it is. We all know that's just white people shit.

But they say I attack people. This is a bald-faced lie. If you look at every single one of my contentious moments with other users on here, they
come at me. I don't attack people. I defend myself. 100% of the time. I do not go after people. That's not my thing, and I have zero interest in the lives of bigots and people who just don't like me for whatever reason.

Earlier in the week, I talked about how much understanding there is for bigotry and hate on the fedi but that isn't extended to it's targets.

The situation is with mas.to is a perfect example of this.

I would urge caution for anyone who interacts with that instance as they've shown a tolerance for bigotry but none for people who are harassed and targeted by it.

We need another sex positive, sex worker instance on the fedi.

We don't have enough smut. Smut levels are far too low.

I, in the nicest possible way, do not care if people like/prefer/need dark mode. Or light mode! It's a real "difference of opinion" situation; your choice doesn't affect anyone else so do what you like!

I just think both should be options, both should be treated as equally important in development, and no one should be mean to the individual people using the other.

It's just weird that the failures to do this, that I see anyway, all seem to be on one side of this dark/light binary.

Show thread

choosing a Mastodon instance/server, special heads-up for mastodon.social users 

Since so many people are joining mastodon.social right now, including some really unpleasant-to-deal-with people, and since mastodon.social doesn't have a lot of hands-on moderation to dissuade the unpleasant folks, some other instances are talking about "defederating" from mastodon.social. That would mean that, if your account is on mastodon.social, folks on other servers would be less likely to get/read your posts.

Show thread

E-mailing me with a technical question is like playing the lottery; if the spoons align you'll get a comprehensive novel in return, but otherwise you'll probably get no response at all and it sits in my "to-respond-to" pile forever

Ah, Elon Musk. He tried to be the Henry Ford of space travel but instead became the Liz Truss of social media.

weird, the type checker keeps rejecting my program and the error message just says "this ain't it chief" or "not the vibe"

twitter fedi meta 

thinking about all the time and effort that was wasted by people trying to "fix" twitter only for it to be bought by an impulsive billionaire who is setting it on fire, and how the real move was to build an alternative for people to come to following it's inevitable collapse. thinking about how this scales to other similarly doomed systems.

Wild how conservative objections to things are either entirely unconnected to reality or are good things actually

Als je tegelijkertijd leest dat Pfizer €21 miljard winst maakt en dat passende zorg voor de burger niet meer gegarandeerd kan worden, is de eenvoudige conclusie dat doorgeslagen marktwerking in de zorg het probleem is.

Hey, I'm looking for a specific fedi instance recommendation

Allows bots (it won't be an annoying bot)
Has a long (>1500) max toot length
Supports text/html toots or similar (not upstream mastodon, but glitch-soc and some other forks are fine)

Show older
Pixietown

Small server part of the pixie.town infrastructure. Registration is closed.