Re: reason why beagle doesnt index

[Andreas Heinz <dashboard a80 net>]

hi there,

thanks for responding. regarding the file enconding i had some talks to 
kevin and i'm just having a look if there is something what i could do.

dropping the message so that it doenst appear in the logs anymore might 
be a good idea, but i don't know if this should be the final solution.

i was worried about the error message and the logs not being indexed 
properly since at first searching for specific data didnt return any 
result. that was, why i thought it was a problem cause of rewriting the 
backend.

bye
Andreas

Joe Shaw wrote:
> Hi,
>
> On Mon, 2006-07-17 at 14:29 +0200, Andreas Heinz wrote:
> > i tested with: mono_external_encoding=iso-8859-1. now it can read the files.
>
> Yeah, the message you saw was actually from Mono, not from Beagle.
> There's nothing we can really do about that message.
>
> Encodings are an issue generally for Beagle, because since you can't
> know what the encoding of a filename is, it's difficult to pass it
> around and index it effective.  It's important that filenames are in
> UTF-8 or else things might not work completely.
>
> > but now i get following errors in the logs:
> > "060717 1402539793 06093 Beagle ERROR: Couldn't write attributes for 
> > /home/crash/.gaim/logs/icq/22013956/116305009/2005-11-08.234921.txt"
> >
> > for me it seems like beagle tries to use EAs which it can't use for my 
> > gaim logs since they are symlinked to a nfs share and on my harddisk 
> > because of xfs. dont know if the symlink can be a problem.
>
> The lack of xattr support shouldn't cause any real problems, although it
> will be quite a bit less efficient.  But data should still be indexed.
>
> > after having run beagle for about a hour, some of the information which 
> > comes out of the gaim logs is reachable from beagle-query. but i don't 
> > know why the gaim logs know make problems.
>
> I'm not sure I understand.  Are you worried about the error message, but
> all of your gaim logs are indexed properly?  That is the intended
> effect; maybe we should bump that warning down from ERROR to WARN.
>
> > could this gaim problem result from rebuilding my gaim index, because of 
> > joe's "rewrite" of the gaim backend?
>
> It shouldn't change anything from the previous backend.
>
> Joe