nnml compaction

[Didier Verna <didier@lrde.epita.fr>]

        Dear folks,

I've just committed a long awaited feature (I think): nnml compaction at the
_backend_ level. Here, compaction means removing the gaps between article
numbers (and actually starting again from article number 1). This gives you
more accurate information (like total number of articles and so on).

No more need for each user to create a new group and move all articles to that
group, or any other manual trickery that we've seen before. To use this
feature, you can type 'G z' on an nnml group in the Group Buffer, or you can
type 'z' on an nnml server in the Server Buffer (All groups in this server
will be compacted).

This feature is fully functional for me, but should probably still be
considered beta. In particular, it correctly preserves marks, but nothing is
written to support the agent or the cache. I don't know these beasts and I
don't even know whether I should be concerned by them. Any expertise on the
matter would be appreciated. Also, there is still a display problem with the
server compaction function: it doesn't update the information in the Group
Buffer. 

I've run it on my main nnml server and it took 4 minutes to compact 25000
messages. This result is totally meaningless because I have no information on
the topology of the gaps I had (I just know there were plenty because of
removed spams). However, it shows that the feature works ;-)

Last thing: I've not updated the Gnus manual yet.


Try it, but backup first :-)

-- 
Didier Verna, didier@lrde.epita.fr, http://www.lrde.epita.fr/~didier

EPITA / LRDE, 14-16 rue Voltaire   Tel.+33 (1) 44 08 01 85
94276 Le Kremlin-Bicêtre, France   Fax.+33 (1) 53 14 59 22   didier@xemacs.org