Bug#1031377: UDD/lintian: Needs to run lintian on all binaries generated from same source
On Thu, 16 Feb 2023 15:05:24 +0100 Lucas Nussbaum wrote: > What could work is: > run lintian on source > for each arch in the packages's architectures (except all) > run lintian on architecture packages + architecture 'all' packages > > But would that solve all issues? I discovered that it does not solve all issues. For example src:rust-bat builds bat. When you run lintian against the dsc and the deb it emits missing-notice-file-for-apache-license for the source package, but when you run it against only the dsc or only the deb it does not emit that tag. That is because the test checks that the NOTICE file from the source package is missing from the binary package. https://lists.debian.org/msgid-search/8062de5e3a1afbe988ce3d5453d211089242ba2a.ca...@debian.org -- bye, pabs https://wiki.debian.org/PaulWise signature.asc Description: This is a digitally signed message part
Bug#1031377: UDD/lintian: Needs to run lintian on all binaries generated from same source
(Adding lintian-ma...@debian.org to Cc for input) On 16/02/23 at 11:18 +0800, Paul Wise wrote: > On Thu, 2023-02-16 at 01:17 +0100, Guillem Jover wrote: > > > it would need to get the list of binary packages for a source and > > lint all of them with the same lintian call. > > The usual way of running lintian after a build checks all binary > packages and the source at the same time. I think UDD should do that. That's not very convenient because UDD/lintian runs lintian for each architecture, and lintian's output does not include the architecture for binary packages, so I just get something like: C: libbsd0: control-tarball-compression-format xz C: libbsd0: control-tarball-compression-format xz C: libbsd0: control-tarball-compression-format xz C: libbsd0: control-tarball-compression-format xz C: libbsd0: control-tarball-compression-format xz C: libbsd0: control-tarball-compression-format xz C: libbsd0: control-tarball-compression-format xz C: libbsd0: control-tarball-compression-format xz C: libbsd0: control-tarball-compression-format xz with no way of linking emmitted tags to a specific architecture... What I could do is: for each arch in the packages's architectures (except all) run lintian on source + architecture packages + architecture 'all' packages ... but that would result in a much longer runtime... What could work is: run lintian on source for each arch in the packages's architectures (except all) run lintian on architecture packages + architecture 'all' packages But would that solve all issues? The best way out is probably to have lintian optionally suffix packages names with architecture, and then do a single lintian run with all binaries for all architectures... Lucas
Bug#1031377: UDD/lintian: Needs to run lintian on all binaries generated from same source
On Thu, 2023-02-16 at 01:17 +0100, Guillem Jover wrote: > it would need to get the list of binary packages for a source and > lint all of them with the same lintian call. The usual way of running lintian after a build checks all binary packages and the source at the same time. I think UDD should do that. -- bye, pabs https://wiki.debian.org/PaulWise signature.asc Description: This is a digitally signed message part
Bug#1031377: UDD/lintian: Needs to run lintian on all binaries generated from same source
Package: qa.debian.org Severity: normal User: qa.debian@packages.debian.org Usertags: udd Hi! There are some lintian tags (I know at least of lacks-unversioned-link-to-shared-library), which require multiple packages to be linted together for them to be emitted. In this case libfooN and libfoo-dev for example. I'm getting unused-override tags as these are not triggering (an example would be libbsd). I guess the lintian runner used by UDD is executing lintian for each .deb independently, because I assume it has no access to the .changes files? If so, to avoid these false positives, it would need to get the list of binary packages for a source and lint all of them with the same lintian call. Thanks, Guillem
