Re: Why file content sniffing sucks

[Ingo Ruhnke <grumbel gmx de>]

Jeffrey Stedfast <fejj ximian com> writes:

> the whole detection on "special" filename is a bit weak. there are
> potentially a huge number of them and it is an unclean hack to have to
> test against them.

Its not an unclean huck, its something that should be a users choice.
All I want is a way to configure this whole mess, not replace one
vodoo-magic with another one. Replacing content detection with
executable-bits alone wouldn't do any good, but simply completly
ignoring them and relying solly on content wouldn't be any good
either.

> let the extension override the content-sniffed mime-type IF AND ONLY IF
> the sniffed content can fit within the mime-type that the extension
> suggests.

This would get wrong for all file types that nautilus doesn't know
anything (quite a lot) and actually detectes them to be something
different (not so much but still quite a few), this is happening right
now for me and will for sure happening in the future.

> actually... come to think of it, content sniffing probably doesn't ever
> get anything wrong other than text types anyway...

It can get wrong for all files that it doesn't know about and for
which the detection code is buggy. Raw binary types are another thing
that can't be detected by the content alone.

> this makes it real easy. 

It still leaves the user with some magic jumping in at a place where
he doesn't know about it and where he can't stop it from jumping in.
Sure it might be better then the current aproach, but without a way to
configure all this properly it will still be a mess.

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