User behavior (i.e. having practically 0 tolerance for any relative slowness) has also influenced this. Youāll see plenty of like forum posts recommending disabling swap on Unix systems to āimprove performanceā when that is almost certainly having the opposite effect overall because the advantages of the default settings (chosen by people who almost certainly have forgotten more about computer performance than any of us will ever know) are really, really good in the vast majority of workload profiles. I work at a data center company, and Iāve written internal documentation that is emphatically trying to convince fellow employees to stop fucking disabling it because itās almost certainly going to do nothing but negative things for our customers and itās still a misconception people have.
I guess the other aspect of this is that writing highly efficient programs is really hard. The modern programming landscape is dominated by composing libraries and packages which may or may not have made good or appropriate decisions. Itās often not practical to compare them on an efficiency level because thereās usually not that much choice and the interfaces and other features are much more significant. It doesnāt matter if package A is 20% more efficient if package B is better maintained and has all the types and functions I want. In practice, the best code is the code you did not write and donāt have to maintain yourself. JavaScript and Node, which is horrifyingly is one of if not the most used language for most things we use nowadays, has an insane fucking package ecosystem. The fragmentation is unbelievably sprawling and itās vulnerable to so many supply chain attacks on top of the numerous inefficiencies that naturally stack up in that sort of environment.
AI is almost certainly going to make this worse. Like even using frontier models, my experience is the generated code is at best approximately functional with a bunch of unnecessary nonsense. A coworker uses it a lot and his consistent position is most of it needs to be rewritten eventually, but itās acceptable in the short term crunch we find ourselves in. I donāt think a lot of vibe code is scrutinized as much, and a lot of metrics point to just an explosion of new projects and code, so much so that Iāve seen ecosystem maintainers say that new submissions canāt even be reviewed by humans anymore. Thereās just too much code and they have to rely on automated AI systems. And again I was saying the stuff made by the frontier models still sucks and is bloated in the aggregate. Not everyone is using frontier models.
But anyway, I admittedly write code thatās optimized enough for the environment it needs to run in. I do find it hard to blame other developers for not giving much of a shit about something being super lean in userland on external systems. As long as itās not going to be noticeable and doesnāt feel slow, itās good enough to ship. If there are 2 ways to do something and one is easy to write but unoptimized, and another is difficult to write but optimized, Iām only choosing the latter if the anticipated usage warrants that optimization. Resource optimization is just so, so low on the totem pole of concerns beyond avoiding negative extremes. I donāt have the time or the patience or the incentive because yeah, most of the time I can just pretend hardware constraints arenāt real.
yeah lmao
I am severely blind but is it saying 11037
no im pretty sure it says leon
Itās definitely saying 11037
Idk could be 11O37
No, no, it says N037.
Iām impressed, only Kyoko from all people could have ever deduced such a hint
okay maybe i dreamed this or made this up but wasnāt there a sulit 11037 thing in an old-fol dangan rp
i cant remember the story but i need to know if im insane now
paging @Arete
+!1n5
im crying literally one page in because im just so happy theres more owl house yall iām So Normal
xdd
but hey nothing at all wrong with being happy about stuff you like
map looks pretty neat though the theming could wear thin. the main changes to it are really in the side areas of lanes in archmotherās side but thereās a bunch of little changes everywhere
so, as far as I can remember, it wasnāt that there was actually any kind of dying message implicating sulit. Rather, in FoL DR1 case 1, there was a large amount of evidence pointing towards sulit (e.g. nearly every other player had an alibi, they thought it had to be the Ultimate Engineer and sulit then claimed Ultimate Engineer (I believe that deduction was actually false and she hadnāt used her talent), etc.)
so, as a joke referencing DR1, Marissa claimed to have found a dying message reading ā+!1n5ā (or something along those lines, Iām deferring to Amelia), but there wasnāt actually any such message
having Chinatown could be a massive advantage for hk


