Shared-mime checking order
Alexander Larsson
alexl at redhat.com
Tue Sep 18 04:07:20 PDT 2007
On Tue, 2007-09-18 at 10:37 +0100, Bastien Nocera wrote:
> On Tue, 2007-09-18 at 08:35 +0200, Alexander Larsson wrote:
> > On Mon, 2007-09-17 at 22:03 +0200, David Faure wrote:
> > >
> > > > We should also make sure that we add all possible conflicting extension
> > > > matches to the database. For instace, if we add "*.pdf" to
> > > > application/x-gzpdf then we could use sniffing to resolve which one it
> > > > is.
> > >
> > > I agree. My unit test for the *.doc conflict-resolving shows that "*.doc" should
> > > be added to the glob patterns for text/plain, for instance.
> > > Do you have access for making changes to the specification and to the shared-mime database, btw? (I don't)
> >
> > I think i have write access, but at the very least i know some people
> > who have that i can badger if if turns our I don't. Have you made any of
> > these changes and would like me to apply them?
>
> I should probably have mentioned that before, but there's now a test
> suite in shared-mime-info CVS courtesy of Matthias Clasen. You'll need
> to have xdgmime checked out and compile at the same level as
> shared-mime-info. Running make in CVS will run the test suite.
>
> All changes to the freedesktop.org.xml.in file should have the an entry
> in the test suite.
>
> I'll update the HACKING file to reflect that.
Hmmm. I started working on a test for this change, but its a bit tricky.
The change doesn't really affect anything in the current tests, one has
to add a textfile.doc and try to detect that as a textfile. However,
text/plain has no *.doc match yet, so that doesn't do anything. And if I
add that to text/plain then the xdgmime code seems to fail due to the
conflict, as it only returns application/octet-stream for the name test.
I guess we need to update the xdgmime matching code with the changes
we're talking about in the mime spec. But we need to make sure everyone
agrees first.
More information about the xdg
mailing list