Thinking about how Mozilla is "pivoting to AI" but DeepSpeech, one of the very few "AI" products you could possibly find a positive use for (pure-local speech recognition), is not only a Mozilla product but so abandoned that you actually have to downgrade to Python 3.9 to run it
Like, I saw a post complaining that everyone simultaneously loves Firefox's pure-local translation library but everyone is also complaining about Firefox "refocusing" on AI, and that is a fair point, but also to me if a company/org announces they are going to focus on "AI" that is a statement that they are only going to be exploring the least useful, most damaging forms of "AI". If they intended to do something good, they'd probably call it something other than "AI"
"You can get the source code to Balatro by just renaming the .exe to .zip and opening it" is so funny to think about. Imagining Elden Ring working like this
@leona ohhh i actually very almost shipped a game like this
or rather the Mermaid Heavy Industries game was written like this and it just didn't get shipped
the trick is that .zip files are allowed to have "ignored" junk data at the beginning and .exe files are allowed to have "ignored" junk data at the end, so you have a polyglot file where to the zip the exe part is "ignored" and to the exe the zip part is "ignored" but then the exe fopen()s itself and treats itself as a zip file
Last night as I was about to fall asleep I had a really good idea for a VR game but unfortunately this happened about 6 years too late for me to actually be in a position to turn the idea into a VR game
So... my game dev project has been all one single main.rs all this time, and now I'm splitting things up. I've gone from 1 to 7 rust sources all of a sudden and I wonder if this is dumb or good. Impossible to say.
@onelson I actually because rust allows inner functions and types keep writing complex programs confined within a single main(). Just using that one function as a namespace
Been trying to figure out—in principle, the reason StackOverflow is CC-BY-SA is to allow for community mirrors, right?
Is there any way that I could just get a total local mirror of StackOverflow and search that locally, instead of having to visit StackOverflow which really doesn't deserve my support?
…Or does SO ban scraping, and the only reason for CC-BY-SA was so StackOverflow could (now that the rugpull's happened) claim the right to block u from delerng posts?? CC for them but not for us?
Suddenly occurring to me that if I set myself up with this https://mastodon.social/@whitequark/112459925601002436 I could set up a personal search method for StackOverflow which, unlike the usual StackOverflow frontend (Google), actually weights questions/answers on how close they are to the present, so I can stop getting answers from 2013 saying "that isn't possible" about things that have been possible since 2018
We hoped branch = BRANCHNAME would make it activate BRANCHNAME on clone. But no, it checks out as "detatched head", which later makes VSCode confused. Can we do better?
The submodules are cloned and the branches are checked out.
In submodule A, the commit which is checked out happens to be branch IMPORTANTBRANCH.
Git then "activates" that branch, so that the subrepo is no longer in "floating head" mode, and commits to this subrepo go to that branch instead of into a weird hole.
Visual Studio Code does not warn you, when you add commits, that you are in detached head mode and that committing is not going to do what you want.
I was trying to make things easier for them, so that they could do clones without having to either remember to go manually activate a bunch of branches or risk a weird detached head related screw up