Attachments (1)
Change History (16)
comment:1 by , 17 months ago
comment:2 by , 17 months ago
I'll try to do this tomorrow morning, have an ALFS build running right now to verify the build
follow-up: 4 comment:3 by , 17 months ago
<r1c0n3r0> many apps using were broken due to pkgconf-2.0.0 and rustc-1.71.1, eg. librsvg, gst-plugin-rs, firefox, thunderbird. https://bpa.st/XCMQ <r1c0n3r0> I upgrade pkgconf from 1.9.5 to 2.0.0, not sure if a clean build has the issue or not.
Not sure (yet) what user has done, but we might test pkgconf-2 with those packages.
comment:4 by , 17 months ago
Replying to thomas:
gst-plugin-rs, firefox, thunderbird. https://bpa.st/XCMQ
<r1c0n3r0> I upgrade pkgconf from 1.9.5 to 2.0.0, not sure if a clean build has the issue
or not.
}}}
Not sure (yet) what user has done, but we might test pkgconf-2 with those packages.
I guess that we'll have to test them. At a first glance it looks like this https://github.com/haskell/cabal/issues/8923. But I am unsure
comment:5 by , 17 months ago
Seems like this commit, https://github.com/pkgconf/pkgconf/commit/a97b75ab2c1d031982c35a4886102413e4ec8eee. It seems logical, so we could either hold out on 1.9.5 (Arch and Debian are both on 1.8 and there are no security fixes pushed) while packages update. Or we could revert the change via a patch. If we use 2.0.0 unpatched, firefox will definitely need a patch because it's the ESR release.
comment:6 by , 17 months ago
We could try something like this for pkgconf-2.0.0:
sed -i '/1330,1336/s|^|//|' cli/main.c
follow-up: 14 comment:7 by , 17 months ago
This is an interesting issue since it causes other regressions, see: https://github.com/pkgconf/pkgconf/issues/310
Most of the packages listed above are rust users, which means that it could've been a problem with the pkg_config crate - but that's just a theory (though it is corroborated with this PR: https://github.com/rust-lang/pkg-config-rs/pull/151/files). Bruce's sed though is probably the best approach since we probably want the --with-path fix. Updating rust crates within existing packages requires calculating the checksum of the crate again and updating Cargo.lock (it checks to make sure nobody's modified the crate for security reasons), so we probably don't want to do that unless we have to (see the bindgen patches for Firefox earlier in the cycle for details on how you'd do that). That would need to be done to every impacted package.
follow-up: 15 comment:8 by , 17 months ago
Yeah I think that should work, The ALFS build boots and all the pkgconf elements files work just as well as before. I'll have a patch soon
by , 17 months ago
Attachment: | pkgconf-2.0.0.patch added |
---|
comment:9 by , 17 months ago
Builds fine on a fresh LFS system, the BLFS packages in question were not tested, but I will see if I can test them later today. I assume that they work because of the sed. Patch is attached
comment:10 by , 17 months ago
I'll use that patch and at the same time bring LFS completely up to date. I should be able to commit in a couple of hours.
comment:11 by , 17 months ago
Fixed at commit 9b7d651a54314b80138f1254a64005e6b7e95e53
Update to xz-5.4.4. Update to wheel-0.41.1 (Python Module). Update to man-pages-6.05.01. Update to linux-6.4.8. Update to iana-etc-20230804. Update to pkgconf 2.0.0.
comment:12 by , 17 months ago
Resolution: | → fixed |
---|---|
Status: | new → closed |
comment:14 by , 17 months ago
Replying to Douglas R. Reno:
This is an interesting issue since it causes other regressions, see: https://github.com/pkgconf/pkgconf/issues/310 Updating rust crates within existing packages requires calculating the checksum of the crate again and updating Cargo.lock (it checks to make sure nobody's modified the crate for security reasons)
FWIW we can also add a patch section info Cargo.toml and it's more "understandable" to human. See the last hunk of thunderbird-102.12.0-consolidated-1.patch for example.
comment:15 by , 17 months ago
Replying to rahul:
Yeah I think that should work, The ALFS build boots and all the pkgconf elements files work just as well as before. I'll have a patch soon
Please try to make a good explanation (LFS is an educational material, not only a distro so the text has to be more accurate than the comments in PKGBUILD or whatever files in other distros) and keep the style (both the XML source and the rendered HTML) consistent with everywhere else in the book.
See r11.3-204-gf4835ed6a.
Changes from 1.9.5 to 2.0.0: