How about we create a home-service called Guix-QoL and we pull the guix keyring part of parcimonie into it (or something similar) and add a Cron job to run "git gc" on all the repositories in ~/.cache/guix/checkouts and add more bits later as we find them. Then the home service can have a series of #t/#f switches to clean up after guix.
#libgit2 #Git
@nathanstocks Tiny dataset - maybe, but I'd have to understand the problem first! On looking at the code (and after your note 3) I'm not actually sure if it's in git loading at all.
Update - as I wrote this, my "run the thing with verbose logging" job finally finished with a more useful error message. And it's not the git code. Sigh. (I'd assumed it was git, becase a previous run without git had succeeded - but turns out that was a red herring. Sorry for the slight, #libgit2 !)
The root problem it turns out is non-utf8 code in a file I was scanning: https://github.com/spring-projects/spring-boot/blob/main/spring-boot-project/spring-boot/src/test/resources/org/springframework/boot/env/test-properties.properties - which broke something else. Need to look at why it broke (which I don't mind) and why the error was so contextless (which I do!)
Hey #rust #rustlang folks - something in my app is causing it to exit with a message "Error: stream did not contain valid UTF-8" - and adding RUST_BACKTRACE=1 gives no backtrace! Suggestions on things I could try next?
Sadly this happens only for one dataset, and it takes 40 minutes of processing to get to the point of failure, so diagnosis is painful.
More context - I suspect that the problem is somewhere in git2-rs
which wraps #libgit2 - I can try narrowing it down with catch_unwind
wrappers, but it's going to take a while (due to the 40 minute feedback loop)