On Mon, 17 Mar 2008, Magnus Hagander wrote:
> We are sucking a lot of this data down to the db on search.postgresql.org
> already. Does it make sense to do it there perhaps?
No, using the database for this sort of thing is so old-school at this
point. Didn't you hear that good web applications abstract away the
database so you don't have to worry about what's in there? The right way
to handle this is to push the entire archive through the client each time
so it can make rich decisions about the data instead. I hear Ruby on
Rails is a good tool for this.
> Is there need for anything more than a unique-messageid-hit? If that's
> all we need, we could easily have an url like
> http://search.postgresql.org/search?msgid=19873987123@foo.com redirect
> to the proper page on archives?
That would be perfect. I'd like to be able to replace my saved mailbox
with a web page containing links to the archives instead, and that would
be easy to do with that support. Bruce's requirements have a similar
mapping job to accomplish, and I could imagine a useful app for that area
that consumed an mbox file and output a page of wiki markup. Why, the
patch queue is practically on a wiki already!
--
* Greg Smith gsmith@gregsmith.com http://www.gregsmith.com Baltimore, MD