Re: Why file content sniffing sucks

[Ingo Ruhnke <grumbel gmx de>]

Elliot Lee <sopwith redhat com> writes:

> e.g.:
> 	For *.txt filenames, always treat them as text/plain
> 	For a 'README' filename, guess based on content.

Why start guessing for README on the content? I haven't meet any
README file in my live that wasn't plain text, and the ones that were
not always had that their .html, .doc, .whatever suffix. Guessing on
content should only be happen, if at all, if there is absolutly no
other way to get the filetype.

After all the filetype determination shouldn't be limited to the
suffix, but better to a globbing pattern or regular expression, after
all if you handle some files of from the Amiga you will have filenames
with the type first in the filename, not last (ie.
mdat.some_music_file).

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