Re: B m completion

[Guido Van Hoecke <Guido.VanHoecke@advalvas.be>]

Kai Großjohann <Kai.Grossjohann@CS.Uni-Dortmund.DE> writes:

> Guido Van Hoecke <Guido.VanHoecke@advalvas.be> writes:
> 
> > However, the drawback is that nnimap/gnus queries my imap server for
> > new mail in each of these folders. This is time consuming and
> > completely useless: I never have any mail delivered to any of these
> > folders.
> 
> Sometimes, nnimap says `checking group foo' for all groups.  This is
> really fast.  At other times, it says `updating info for group foo'
> for all groups.  This is really slow.  If you see the latter, I think
> it would be best for you to make it so that nnimap says `checking'
> instead.  I think you will observe an order of magnitude in speed
> increase, much more than could be achieved otherwise.

Yes, I have now observed this, and indeed the difference is an order
of magnitude. I used to call (gnus 5). Now I just call (gnus) and the
initial phase uses 'checking...' rather than 'updating...'. Boy this
is fast.

However, no matter what I do, the next time around, e.g. by keing 'g'
in the group buffer, or M-x gnus again, it always reverts to 'updating...'

> 
> I'm not sure what needs to be done to make it so, but I think the
> problem is the select method. I have (manually!) set the select method
> of all my nnimap groups to "nnimap:" (no parentheses!), and this
> seemed to work well.  Maybe you need "nnimap+foo:", though.

I have manually edited my select method, and set them to "nnimap:se"
for all my imap groups. But this does not make any difference.

> 
> And I find that nnimap switches from `checking' to `updating info'
> when doing M-x gnus-no-server RET as well as when invoking M-x gnus
> RET with a specific level.  Oh, well.
> 
> But maybe this discussion should be in the nnimap list...

Even if somebody can help me with a mechanism to stick to the
'checking' method, I still would like to be able to define that some
groups never need a 'get-new-mail': such a group parameter would still
provide a faster method than the 'checking' approach.

Guido.