Previously, the core would fallback to other demux plugins which was
inefficient and error-prone.

A plugin can use this if it matched but detected a non-recoverable error
while inside the open callback. Help yourself if you want a "better"
error code.

exit() cannot return negative numbers (the high-order bits are ignored).
And anyway, that's not a meanigful place for VLC error codes. It is also
questionable whether we should exit() from LibVLC anyway...

This is consistent with existing practice inside and outside of VLC.
We really don't need a separate error code for that.

Nothing can really kill VLC when browsing plugins, since nothing has
started.

This should be more clearer to user what column is used to sort (eg if
it's hidden, then hide sortIndicator also).

If you setSortingEnabled after setting model, it doesn't call sorting,
otherway around it does. Missed thatone on previous try.

This is inefficient. It could be optimized a lot with
epoll/timerfd on Linux and kqueue/kevent on BSD. Unfortunately, the
POSIX timer API is impossible to use in a thread-safe manner, other
than with signals.

Note that this can only be used safely in the core. In a plug-in, it
would introduce a race whereby dlclose() unmaps the code segment that
a detached thread is running (this was discussed over a year ago).
For that, we'd need something a bit more involved along the lines of
Win32's FreeLibraryAndExitThread().

This is never used currently. Removing it simplifies the next commits.

macosx: prepare the switch from dyld to 'Dynamic Loader Compatibility Functions' as partially intended by [7147338b7439f7f4a5d6d9c86e8d5d7a8501be4d]

Using Apple's legacy object file image functions around NSModule is strongly discouraged and won't work in the future at all.

This reverts commit 75b34a6c.

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

This reverts commit 3266085e.
It triggers a crash whenever sorting manually.

On all my linux PC (debian sid and lenny) it has really bad side effects,
mostly when seeking and at startup up (slow or even blocking), and makes
sshfs unusable.

fontconfig doesn't really seem to set that result variable correctly
in allmost all cases, only few codepaths that it's setted to NoMatch,
but it doesn't get setted on valid paths or configerrors or any of
thosecases. Thanks to Laurent Aimar for this one.