Re: Scope of nnmail-split-methods

[Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann)]

Nevin Kapur <nevin@jhu.edu> writes:

> Yes, this works. So if I had an nnml group JHU and an nnfolder group
> JHU, (I use both backends) where would the mail end up? I could not
> figure this out from the documentation.

Ah, that one is a good question.  I also had this question.  After
some chatting (well, mail-exchange) with Lars, it seems that the
procedure is as follows: Gnus goes through its servers in some order,
and the first server that can receive mail gets all the mail.  The
only thing that's left to do is to find out what order exactly that
is.  I'm pretty sure that first the native server is checked (ie,
gnus-select-method), and then the secondary servers are checked in the
order given in the variable gnus-secondary-select-methods.  I think
that after this come the foreign servers, presumably in the order
given in the Server buffer.

It is conceivable that Lars will change the mail-source stuff such
that you can define different sources and maybe different split
methods for different servers, such that the first server which can
receive mail does not necessarily receive all of it.

PS: I'm noticing that I didn't define what it means for a server to be
able to receive mail.  Well, it is obvious that an nntp server cannot
receive mail, neither can nndir and nneething.  But I'm not sure
about, say, an nnfolder server with the nnfolder-get-new-mail backend
parameter set to nil.  I can think of two things which might happen.
Either this nnfolder backend is recognized as `cannot receive mail',
or the nnfolder backend is recognized as `can receive mail' (because
nnfolder backends can receive mail) but receiving mail has been turned
off, so in the end no backend gets any mail at all.  Dunno if
something different happens if the variable nnmfolder-get-new-mail is
set globally.  What's the story?

kai
-- 
A preposition is not a good thing to end a sentence with.