Re: Maildir support status


On Thu, 2005-07-28 at 00:34 -0400, D Bera wrote:
> Where exactly is the difficulty in creating a MaildirIndexable and
> MailCrawler for indexing maildir files ? In my opinion they can be
> just wrappers around the FilterMail filter. That would index them as
> mails and the hits would behave as mails too.

A specialized backend (like the EvolutionMailQueryable) makes sense if
where you are getting the files from is a data source.  For example, a
KMailQueryable makes a lot of sense, because there is additional
information that you can get from a KMail data source, like the names of
folders.  It's also necessary for finding out what directories actually
have the mail files to crawl.

The problems with maildir files showing up and behaving as files rather
than mails are just bugs in the tiles and Best.  That's where those
should be fixed.

> I made a rough attempt at doing that and created a simple
> MaildirQueryable. It seems to work for me. Note that it is quite basic
> (and no crawler support yet) and assumes that your mail directory is
> ~/.Mail. If anybody wants to give it a try, get it from
> If this is not the write method, then it would be helpful if somebody
> can explain the proper steps or problems with this approach, then it
> will be extremely helpful.

It looks okay from a first glance, but there's not really any
interesting logic here.

        * You've hardcoded ~/.Mail.  Where does that come from?  Could
        it be different?  How do other mail folders work?
        * Is there extra information that can be extracted from the data
        source that isn't handled by the MailFilter?  Things like folder
        name, account name, etc.  To see a good distinction, look at the
        differences between beagled/EvolutionMailDriver/*.cs and
        * If there isn't anything to add, does it make more sense to
        just tell the file system backend to crawl this directory?

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]