Quoting Peter Michael Green (2025-08-03 04:37:31) > > > The easiest way to unbreak this loop (and we have to do this) is IMO to > > > temporarily disable the log feature of rand. This avoids rand pulling in > > > log, breaking the cycle. Then at least ahash should built. I will do a > > > quick testrun of both and report back here. > > > > I don't like the "temporary" part of that plan. > An alternative soloution is to at least reduce the problem is to > upgrade hashbrown to 0.15, hashbrown 0.15 changes the > default hasher to foldhash and no longer has a dependency > on ahash (you can still use it with ahash through the > "hasher" trait if you want). > > foldhash has no cargo dependencies, build-dependencies or > dev-dependencies. > > I've uploaded hashbrown 0.15 and it's most important reverse > dependency indexmap to experimental and with that and some > other dependency tweaks, I was able to get rust-uuid installed > in my experimental chroot without needing rust-ahash. > > unfortunately, rust-ahash also build-depends on criterion, > which depends on ciborium, which depends on ciborium-ll, > which depends on half, which depends on rkyv which depends > on ahash > > Perhaps disabling the ahash feature in rkyv would make sense, > I don't think anything in Debian uses it. > > Anyway, I need to get some sleep, will have a further look at > this sometime.
I have tried to avoid benchmarking tests to get rid of the dependency on criterion, but then another no-panic test mysteriously (to me) fails. I have pushed my changes to https://salsa.debian.org/debian/rust-ahash and would appreciate if someone more clueful than me could try make sense of why no-panic test fails. - Jonas -- * Jonas Smedegaard - idealist & Internet-arkitekt * Tlf.: +45 40843136 Website: http://dr.jones.dk/ * Sponsorship: https://ko-fi.com/drjones [x] quote me freely [ ] ask before reusing [ ] keep private
signature.asc
Description: signature
