More on mime wierdness

["David Rowntree" <therealdaverowntree googlemail com>]

All,  the last thread about mime wrt jpegs caused me to look at my index.  It would appear that at least _my_ system (standard suse10.1) is not indexing compressed windows media. 

If I take a text file, and create pdf, ppt, doc files from it using openoffice (or whatever) Then let the indexer run, perform a search on the text, I get hits for each media type, as expected.
Now if I compress all the files with gzip and then re-search, I only get hits on the pdf.gz file.  Looking at the logs, and running the beagle-extract-content program reports that there are no filters for the ppt.gz, 
doc.gz files as the mime type of the child indexables has no filter.  It seems to be looking for structured-storage instead of vspowerpoint, for example. (sorry can't remember exact details..more to follow) and fails.

The mime type is correct for uncompressed files, but not for child indexables - does this point to a difference in mime-detection for child indexibles?  Is the mime detection done exclusively by beagle or is it my system setup that is inconsistant?  Where in my system config is the mime 'database' ?

I will produce some test case examples, showing the logs, in a day or two when I have chance!

 (incidently this is tried on 0.2.14, I was waiting for 0.2.16 because of the mono version requirement meant I had to do some major upgrading - which I'm glad I didn't :)

Kind Regards,
Dave.