[Rpm-maint] [rpm-software-management/rpm] configure.ac: add linux-gnux32 variant to triplet handling (PR #2143)

[Michal Domonkos]

> > Let's try a different angle: should rpm agree to be (cross) compiled for a target that it doesn't know about? The patch could perhaps be reworked, so that rpm doesn't reject triples that don't match a hardcoded list.
> 
> The answer is no: it should always know the targets, because that data is important for creating a correct RPM with the correct architecture information.

Yet, we used to support compilation for targets we otherwise don't know about until not too long ago (4.17.1). Apparently, some folks (Yocto in this case) have successfully done this until we broke it for no apparent reason with commit 1cdb72ae48b7ba689c5c79118f4f0c1b4ffe6b7c.

Installing and creating RPMs are two different things and it seems like there's no reason why we couldn't allow the former without the latter, although in that case, we should somehow prevent the build process from creating `rpmbuild`, shouldn't we?

Either way, I don't think it makes sense to specifically whitelist `linux-gnux32`. Instead, we should either completely revert 1cdb72ae48b7ba689c5c79118f4f0c1b4ffe6b7c or just "own it" and let these edge cases be dealt with by patching `configure.ac` in their environment.

-- 
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/pull/2143#issuecomment-1219682790
You are receiving this because you are subscribed to this thread.

Message ID: <rpm-software-management/rpm/pull/2143/c1219682790 at github.com>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rpm.org/pipermail/rpm-maint/attachments/20220818/08e1d640/attachment-0001.html>