urbeker

The reason I love jujutsu so much.

I think the main mainters are very busy at the moment which is why a few big features got merged immediately after this release. They want some time to let them get tested by people on the edge as they didn't have time to iron everything out themselves.

I work with python in a professional capacity and pip is really pretty bad. It's widely accepted. But that's just because it came first so other package managers learned from its mistakes. Lots of wrappers exists around pip to try and make some of the pain go away like poetry but they just paper over the issue.

Things like all packages existing in the same namespace so your dependency graph can only have one copy of a library. Just awful compared to more modern alternatives. It makes resolving a dependency graph take forever for one thing and can make upgrading anything a real pain. Also the way installing a package can compile arbitrary c code is nasty, can cause some really horrible issues.

This whole thing is incredibly immature and seems to be blown up to an absurd degree.

Did ThePHD unfairly have their hopes raised about giving a keynote speech? Sure. Was this done out of malice? After - for some reason - thousands of words of aired dirty laundry I can't see any indication it was. I can't see how hyper-analyzing everyone's testimonies and expecting people to have a perfect memory of things they never thought they would have to post-hoc justify helps anyone. Is it really reasonable to expect people that get into highly technical positions based on technical merit to also be perfect organisers and have flawless interpersonal skills?

If there was some catalog of people being offered keynotes or other prestigious things and having them taken away due to spite I could see this being worth spending time over but as a one-off this all just seems absurd. Unless of course there was some key detail I've missed in the half-dozen blogs posts that makes everyone involved terrible people?

Neat, a way of ensuring zero panics in at least your own code is really cool. I wonder how deep the lint goes.

OK so looking at that PR the reason you aren't getting any errors is because it looks like one of the dependencies is trying to load python and thats where the error occurs. Rust would normally give you a pretty good stack trace but because this is happening in external code it can't and you won't fix it in the av1an codebase. vscode likely has its path setup differently allowing it to find python. I would double check the paths inside the vscode terminal and wherever you are running it.

Dynamic libs in rust are supported but only through a c abi and it's a bit complicated.

Looking at av1an I couldn't see anything showing it is dynamicly loading anything, it looks to statically link to ffmpeg according to the build instructions.