Michael Monnerie wrote:
> 
> Seems to me the problem is the relation of physmessage -> headervalue -> 
> headername
> If that would be physmessage -> headername -> headervalue, there would
> be a dramatic speedup, because the planner could lookup all 
> headername='message-id' before having to look at the headervalues. This is 
> especially true when you look at the number of rows:

There is no direct relation between headernames and physmessages. So no, I don't
see how that could be changed easily. If you can whip up a query that improves
the situation please share it.

> 
> select count(*) from dbmail_headername;
>   2430
> dbmail=> select count(*) from dbmail_headervalue;
>  4979350
> 
> Any chance this could be changed? Or are there many lookups that
> search first on headervalue and only afterwards on headername?

Yes, that is the current situation where headervalues have a direct relation
with a physmessage. Like we discussed in the 'compressing header info' thread,
that will be changed during the 2.3 cycle.

-- 
  ________________________________________________________________
  Paul Stevens                                      paul at nfg.nl
  NET FACILITIES GROUP                     GPG/PGP: 1024D/11F8CD31
  The Netherlands________________________________http://www.nfg.nl
_______________________________________________
DBmail mailing list
[email protected]
https://mailman.fastxs.nl/mailman/listinfo/dbmail

Reply via email to