Hi,

A personal note first, since I'm a fairly new face here. I've spent the last 
several days on this quite intensively, and I've found it genuinely intricate, 
far more so than I expected when I started. I'm still inexperienced with 
debcargo and Debian packaging, so I've tried hard to follow the team's 
conventions and to actually understand why each step is what it is, rather than 
just going through the motions. I hope I've gotten it right, and I'd genuinely 
welcome corrections where I haven't. My motivation is twofold: I have a 
downstream package that needs ruzstd 0.8, but I also want to learn to 
contribute here properly and durably, not just push one thing through.

What I've prepared, sbuild-verified on unstable (unshare) and 
installability-checked against the archive, is a single debcargo-conf change 
covering the whole stack plus every reverse dependency in the archive that 
hard-pins an old version (this also covers #1125883, ruzstd 0.8):

- Stack updated to ruzstd 0.8.3, gimli 0.32.3, object 0.37.3, addr2line 0.25.1, 
backtrace 0.3.76
- Reverse deps relaxed to the new versions: aya, aya-obj, implib, elfx86exts 
(object 0.37); framehop (gimli 0.32); xh (ruzstd 0.8)

Two things worth flagging from the work:

- backtrace 0.3.76 wants cpp_demangle 0.5, which isn't in the archive. Since 
addr2line still uses 0.4, I relaxed backtrace back to cpp_demangle 0.4 rather 
than pull a new crate into the transition.
- xh needed a one-line source change: ruzstd 0.8 moved 
FrameDecoder/StreamingDecoder into the decoding module.

The floaters (wasmtime, walrus, sccache, espflash, repro-env, dwarf2sources) 
build-depend on the unversioned virtuals, so they should only need a give-back 
at migration.

I've opened this as MR !1168 in debcargo-conf 
(https://salsa.debian.org/rust-team/debcargo-conf/-/merge_requests/1168). I 
notice the request has been open for a while and, from what I could find, no 
one else is currently working on the stack. I can't upload, so it would need a 
sponsor, and I'm happy to pass the work on in whatever way proves most amenable.

Greg Lamberson

Reply via email to