Re: new mime detection approach

[Ingo Ruhnke <grumbel gmx de>]

Sean Middleditch <elanthis awesomeplay com> writes:

>> It already fails today for some Gimp files, so I have pretty much zero
>> hope that it won't lead to missdetection with much less common files
>> or new file types yet to come.
>
> Then the detection should be fixed.  There is absolutely nothing wrong
> with the fundamental method of sniffing - there are just bugs.  Bugs
> happen, bugs get fixed, everyone dances joyously, etc.  ~,^

As already mentioned, these are not 'just bugs'. These are problems
that Nautilus will never be able to solve with sniffing. Nautilus will
never know all file types on this planet, and even if it would the
detection would still go wrong every now and then. After all a file is
just a stream of bytes, how you interpret them is a users decision,
there is nothing like a filetype magically associated with a random
stream of bytes. Sure some magic-bytes at the start might give a
pretty good guess, but if my magic-byte-less filetype just happens to
have the same bytes as the magic-bytes of another type the file
detection will go wrong, no matter how many hours of bug-fixing will
be done.

Sniffing is just a good guess on what the user might want to do with
the file, nothing more and it should be threated as such, not as a
100% sure way to get the mime-type of the file or as a replacement of
the mime-type stored in a EA.

-- 
WWW:      http://pingus.seul.org/~grumbel/ 
JabberID: grumbel jabber org 
ICQ:      59461927