Re: Suggestion for file type detection approach

[Stuart Gilbert <stu better domain name>]

> > The suffix should be more authoritive than the content, at least in
> > some cases, ie when I name a .html, .xml file or whatever .txt, I want
> > it to be handled as plain text, not launch mozilla once I click on it.
>
>
> .html/.xml files should be handled as text/plain? Hmmm... It seems you 
> are confusing proper behaviour and the way *you* want things to work to 
> fit your habits.

Surely the way *you* want things to work should be configurable 
somewhere though, without editing the source code and recompiling.
The way one person wants something to work is probably also the way a 
large percentage of people using Nautilus want it to work.
Both ways are good, it should be an option with some sort of 
configuration. Perhaps a per-directory setting, or per-suffix.

Only scan the contents of files which claim to be jpegs or mp3s perhaps.

I guess you could just alter the default application for .xml files to 
sort out any xml file problems by changing it to an editor instead of a 
browser. The options for that are already present. That doesn't really 
solve any big problems though, just the personal preference for xml files.

I still stick with the ideas I posted in my previous message about 
having a user specify suffix file-type resolution in certain directories 
and relying on content-based file-type resolution by default.

Perhaps it could depend on the amount of files in a directory though, 
that could be useful.
If there are upwards of 200 files in a directory then rely on suffixes 
unless prompted to scan by content by the user.
The large majority of files are suffixed correctly, in my experience.


Stuart Gilbert.