michal at harddata.com
Fri Mar 14 19:24:35 UTC 2008
On Fri, Mar 14, 2008 at 12:19:44PM -0600, Kevin DeKorte wrote:
> But I do think that it is a little odd that xulrunner was extracted out
> of the firefox/mozilla/epiphany packages so that you could use it as a
> base for other applications and yet you cannot have the x86_64 and i386
> versions of it installed so that if you have an app that you can't
> compile it can still use the common libs.
It appears that packaging gets in the way.
/usr/bin/firefox is a shell script, which is common to x86_64 and
i386 packages, and which finds "real" binaries to run. Those reside
in separate directories for i386 and x86_64 so there is no conflict.
There are few other common files as well but they are really common
so the same principle applies.
With xulrunner /usr/bin/xulrunner ends up as a symlink to
/usr/lib/xulrunner-1.9pre/xulrunner respectively so you have a
conflict right there. Actually a target of this symlink is also a
shell script and at the first glance there is nothing there
architecture dependent; but even if I missed something then
/usr/bin/xulrunner can be trivially replaced by a script common to
both architectures and doing 'exec ...' on what now is a
corresponding symlink target.
> Maybe we need a xulrunner-libs (where you can have the x86_64 and i386
> packages installed in parallel) and xulrunner (only one) packages?
It does not even look that this is needed but packages has to be put
together in a slightly different way (and/or the current xulrunner
script possibly slightly adjusted). Can you file a bugzilla asking
for that? It looks that you may have some uses for such layout.
More information about the test