> %install should run with a read-only build directory
I don't think this is going to work. E.g. autotoolz-based systems (something in
the autotools, automake, libconf stack) do final preparation steps in the
install target. I think this is inelegant, but not really "wrong". Old meson
versions had a buglet in the i18n module where the po file would be generated
not during build but during installation. But more widely, tools write
installation logs into the build directory. Meson does, I think various Python
tools do (pip?), etc. Anything that gives an "uninstall" command needs to put
the information somewhere.
> %check should run with read-only buildroot to prevent tests from affecting
> packaged content.
People were asking about this a lot in #3010. My motivation for this: build
hygiene and reproducibility. The `%check` section is optional and can be
skipped with `--nocheck` or `--without tests`. The result of a build that
skipped checks should be identical, which would break if anything in `%check`
touches `%{buildroot}`. If `%{buildroot}` is made readonly, we know that we can
skip checks safely and save time. For example, when doing build reproducibility
checks, I'd skip tests, because we're not interested in their result at all,
but we can do that safely only if we are sure that they don't influence package
contents.
> It probably does need writable build-dir because those tests do need to write
> someplace
And same with install: I have seen various sources generate stuff needed for
tests in check targets, not build. In summary, I think that in practice all
phases must be given write access to the build directory.
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/issues/3050#issuecomment-2065876550
You are receiving this because you are subscribed to this thread.
Message ID: ___
Rpm-maint mailing list
Rpm-maint@lists.rpm.org
http://lists.rpm.org/mailman/listinfo/rpm-maint
