[Libreoffice] [REVIEWED][REVIEW 3-5, 3-5-0] The type-detection of extension-less documents is broken

[Caolán McNamara]

On Thu, 2012-01-26 at 14:27 +0000, Michael Meeks wrote:
> On Thu, 2012-01-26 at 14:40 +0100, Fridrich Strba wrote:
> > http://cgit.freedesktop.org/libreoffice/core/commit/?id=0d9aa8cce792d69d5147a2230b61b1cbdc089206
> > 
> > fixes a bug where the catch-all type detector was called first in the 
> > list of type detectors when an extension-less document was passed. This 
> > caused quite a number of macintosh documents to be mis-detected as 
> > encoded text.
> 
> 	Gosh; I'd rather like to know why the explicit code to sort it to the
> front was there in the first instance.

Pushed to 3-5 anyway. FWIW I put the original sort code in, my intent
was to try the "known to claim to open everything" typedetector last and
everything else first.

In an ideal world the filter config would have a priority or some-such
field (like shared-mime-info has) and typedetection would have in
priority order with the .csv and "last-ditch load it as text" filters as
the lowest.

C.