Re: Suggestion for file type detection approach

[Geoffrey <esoteric 3times25 net>]

Peter Lundqvist wrote:
> On Thu, 25 Dec 2003, Fabio Gomes wrote:
>
> > Em Qui, 2003-12-25 Ã s 14:55, iain escreveu:
> >
> > > Which is more common
> > > a) Files which are badly named?
> > > b) Files which are funny and nautilus sniffs the wrong thing?
> >
> > Your b) is probably more common.
>
>
> Whichever is most common is hard to say.
> Which causes the most grief, however, is b).
>
> These faults occur when somone missnames a file.

This is based on the assumption that all files would have a suffix, 
which is not the case historically in UNIX OSs.  After all, it is 
/usr/bin/vi, not /usr/bin/vi.exe

> The lower label (if present) shoul perhaps be grayed out to signal that
> this is possibly wrong. It would be healthier to assume that one thing is
> the most probable to be correct - I'd prefer that to be suffixes.

I would suggest that we don't want two guesses.  If we're not sure, take 
a best guess and add a '?' to the type indicator.

Otherwise, where does it stop?

This file has a name that says it's a jpeg
This files content appears to indicate that it's actually text
This file has a tag which now makes me think it's html...

--
Until later, Geoffrey	esoteric 3times25 net

Building secure systems inspite of Microsoft