[mime-info database] inode types

Jaap Karssenberg j.g.karssenberg at student.utwente.nl
Sun Aug 24 14:53:55 EEST 2003


On Sun, 24 Aug 2003 12:06:42 +0100 Thomas Leonard wrote:
: On Sat, Aug 23, 2003 at 10:51:50PM +0200, Jaap Karssenberg wrote:
: [...]
: > One could also argue that when the 'inode/file' type exist this
: > should be the default instead of 'text/plain', because it is a more
: > accurate description for a file without a determined mime-type.
: 
: What's the difference between inode/file and application/octet-stream?
: They seem to be exactly the same thing (ie, any regular file). What
: would you use inode/file for that you can't do with
: application/octet-stream?

I won't start arguing about naming conventions, although I'm tempted.
The difference is I think that 'application/octed-stream' can be
absolutely anything, while 'inode/file' appeals to the fact that it is
an existing inode in the local filesystem. This might make a difference
for certain application frame works.

If you don't want to use it as a default - after all
'application/octed-stream' is around all ready for some time - consider
adding 'inode/file' anyway for completeness sake.

Whether the default is application/octet-stream or inode/file, I think
the default should always be used when the mimetype could not be
determined, and the 'text/plain' heuristic should be removed as default
method. It prevents me from configuring different actions for plain text
files and undetermined files. Also, the spec is fuzzy about the
implementation of this heuristic. The heuristic would have a certain
value as a separate library call.

-- 
   )   (       Jaap Karssenberg || Pardus [Larus]
   :   :       http://pardus-larus.student.utwente.nl/~pardus
 )  \ /  (    
 ",.*'*.,"     " Where computers are concerned _always_ be paranoid "



More information about the xdg mailing list