Re: Why file content sniffing sucks

[Ingo Ruhnke <grumbel gmx de>]

Charles Goodwin <charlie xwt org> writes:

> Well there's a few implicit issues here.  You're saying 'determination
> by content is rubbish' but in the same breadth 'recognise the file
> extension' when it is globally recognised that the whole .ext approach
> is one of Window's greatest flaws.

Having the file-extension stored in the filename itself isn't really
the best way to do it, having a file-type-property that could exist
without clobbering the name would quite a bit better, but Gnomes
'guess on content' way of doing things so far is the worst method I
have ever seen to get the filetype.

> The 'mime-magic' should not be recognising this as an html file. 
> Period.  It's a bug, log it in Gnome's bugzilla.  That's far more
> helpful than a rant like this.

Its not a bug, it just shows that file-content and the meaning of a
file are two different things. Every html file is also a plain text
file, however how it should be used shouldn't be detected by the
computer playing the guessing game, but on the file-type that the user
gave the file (via the suffix due to the lack of a better way to store
it).

> I'm not sure why you're complaining rather than just use something
> that you prefer (xfe) whilst keeping Nautilus in reserve for
> 'content rich' browsing.

I haven't seen a single case where 'content rich' browsing worked
better than plain simple suffix detection, but I have seen a lot of
cases where 'content detection' resulted in nautilus becoming pretty
much unusable.

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