On Fri, 2013-08-23 at 05:40 -0700, Adam Williamson wrote:
> However, to do this, I run into the fucking
> convert-a-directory-to-a-symlink old chestnut, and RPM/yum just isn't
> having it. Following the breadcrumbs all over this list and Bugzilla I
> came up with this %pretrans:
>
> %pretrans -p <lua>
> -- changed dir to symlink in 0.9.3-1, workaround RPM issues
> path = "%{roundcubedir}/program/js/tiny_mce"
> if posix.readlink(path) then
> os.remove(path)
> end
>
> But yum/rpm just won't wear it. 'yum update mynewroundcubepackage.rpm'
> fails with a file conflict just like it does if there's no %pretrans at
> all. I was able to test my unbundling simply by manually removing and
> reinstalling the roundcube package, but I can't push it out anywhere
> unless I can get a %pretrans which yum/rpm will be happy with. Anyone?
> https://bugzilla.redhat.com/show_bug.cgi?id=975909 may be involved, I
> guess.
>
> roundcubemail ships tinymce
> as /usr/share/roundcubemail/program/js/tiny_mce and hooks into it rather
> untidily; it seems rather cleaner to just do 'ln
> -s /usr/share/tinymce/jscripts/tiny_mce
> %{buildroot}%{roundcubedir}/program/js/tiny_mce' in the spec rather than
> write a messy patch for roundcube to change where it looks for tinymce,
> which would probably need constant updating.
Just to cover my ass, this kind of symlinking is explicitly allowed by
the draft new JavaScript policy:
"Regardless, web applications may want to make subdirectories of
%{_jsdir} available under their own directory via aliases or symlinks
for compatibility purposes or to eliminate needless deviation from
upstream."
https://fedoraproject.org/wiki/User:Patches/PackagingDrafts/JavaScript
so I reckon it's fine to do it, so damnit yum, let me.
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net
--
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxxx
https://admin.fedoraproject.org/mailman/listinfo/devel
Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
