Comment # 16 on bug 899627 from Kristyna Streitova

(In reply to Jean Delvare from comment #14)
> (In reply to Kristyna Streitova from comment #13)
> > Also, the following packages needed to be switched to "BuildRequires:
> > p7zip-full" because rpm uses 7za when extracting .7z tarball:
> 
> Looks like the wrong way to fix it. It is rpm which should be changed to
> call 7zr instead of 7za, so that p7zip-full isn't needed.

Yes, that's true. But I don't think that rpm upstream would change it to 7zr
because p7zip binaries separation is done differently across distributions (now
we have the same approach as Debian at least). Of course, we can patch it in
rpm on our own but I'm not sure if it's worth to have a downstream patch just
because ~4 packages are using .7z sources. However, I can discuss it with rpm
maintainer.

> Same holds for all 
> other packages with direct dependency. The whole point of splitting the
> extra binaries to p7zip-full was that 99% of the users could live without it
> and get a smaller base install. If 15 packages trigger the installation of
> p7zip-full directly or indirectly, including the very popular ImageMagick,
> then the whole effort is ruined :(

I understand what you mean. But I believe that the problem is that packages
that require p7zip often "abuse" 7za/7z binaries for cases where 7zr is
completely enough. Usually, they only need to extract .7z files and one really
doesn't need to use 7za binary for that. Good news is that some of them already
realized this and they provide proper support (that respects different
approaches in different distros), which means to search for 7zr, then for 7za
and then for 7z. 
Another part of depending packages use 7za/7z correctly because they use it for
extracting also zip, kmz, lzma, jar, cpio, arj, rar, swf, lha, iso, cab, deb,
rpm and other archives. In that case, requiring p7zip-full looks very
reasonable to me.