Re: beagle-search & beagle-query results mismatch

[Joe Shaw <joeshaw novell com>]

Hi,

On Wed, 2006-08-02 at 17:15 +0100, Paddy Spencer wrote:
> I don't know for that query which is the more correct, which in itself
> is a problem. Is there a way I can open each of the URIs returned by
> beagle-query?

Not really easily, because the actions tend to depend on the hit type
more than on the URIs themselves.  (There is no default handler for
email:// URIs, for example; we have to call evolution explicitly.)

If you can narrow it down to a fairly small subset, it should be
straightforward to verify it by hand.  (But see below first.)

> Oh, does the #n mean an attachment, then? I didn't realise that.

Generally it means that it's a child document of some larger document.
For emails, this means attachments.

> This query was just an example, though - the anomaly I originally posted
> is still present: a search for "lumberjack" gives no results in
> beagle-search and two in beagle-query:
> 
> [pspencer lt2-pspencer rbos]$ beagle-query lumberjack
> email://local local/Personal Folders/Azul/UK;uid=1337#0
> email://local local/Sent;uid=2640#0
> 
> Furthermore, those shouldn't be given as attachments, as there isn't
> one, but they're in the HTML part of the mail that Evolution saves - ie
> it doesn't store locally a plain text version. So is the problem here
> that beagle thinks it's an attachment but beagle-search can't find one?
> I can forward you the email source to look at, if it helps.

Evolution should be storing the entire email, which would contain (in
almost all cases) both a plain text and an HTML part of the mail.  From
the MIME standpoint they are both "attachments" although most mail
readers (and Beagle) try to be smart about it.

It looks like there is some larger bug here with the way (at least
these) HTML mails are being handled.  Can you save a copy of one of
those emails and send it to me?  It would also be helpful to have the
output of "beagle-query --verbose lumberjack" to make sure that it's not
something that is apart of the HTML metadata and not the plain text
version.

Thanks,
Joe