Manual weekly update to cargo-rbmt (to 0.5.1)#1005
Conversation
e3f7400 to
29c5824
Compare
|
Drafted this pr to wait on 0.5.1 benches fix for rbmt, which would render a commit in this PR obsolete. |
9c3b153 to
b505b70
Compare
|
Hey @nyonson, I also had a problem with rbmt 0.5.0 about disk space. So the CI failure on 03891c4 was fixed with b505b70, in which I reduced the log volume with log discarding configs. Is this correct? Or is there some objection to doing it? |
|
I don't know much about those log settings, but after a brief reading it seems sensible to me. I can't think off the top of my head what changed in |
|
I believe |
530bc3e to
035ef24
Compare
|
Great point, I kept only the one essential flag. Realized that the problem actually is not related to rbmt 0.5.0, but a pre-existing issue. It induced flaky CI and just happened to affect my previous run. I kept it included on this PR because it is small and targeted. |
ab89486 to
c7196b1
Compare
satsfy
changed the title
rbmt 0.5.0 removed the standalone docsrs subcommand and folded docs.rs validation into docs, which manages its own toolchains. Keep a single Docs job running cargo rbmt docs and drop Docsrs.
Cargo rbmt compiles the full feature matrix into one target dir, which fills up runner disk. Drop debuginfo logs to save disk space. This is a fix for a pre-existing CI flake.
|
Force push removed the docrs job as discussed in rust-bitcoin/rust-bitcoin#6547. Now bumping to rbmt 0.5.1. |
2 participants
Supersedes #1003
The bump is straight to 0.5.1 to leverage a bug fix in the minor version.
The new rbmt 0.5.0 changeset modifies subcommands, breaking two CI jobs:
docsrsbecamedocs, which by default isdocs.rsmode and accepts--no-docsrs.runnow injects-p <name>for every workspace member.descriptor-fuzzusesold_miniscriptdep.