[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Pan-users] Multiple Server support for PAN: Need a spec
From: |
Charles Kerr |
Subject: |
Re: [Pan-users] Multiple Server support for PAN: Need a spec |
Date: |
Mon, 21 Oct 2002 09:33:38 -0700 |
User-agent: |
Mutt/1.3.20i |
On Mon, Oct 21, 2002 at 08:57:28AM -0500, Gorwood, Steve wrote:
> 1) Assign each server a number
> 2) Maintain a single composite list of newsgroups. Each entry
> in the list of newsgroups will have an added 32 bit field which
> will have a bit set for each server which contains that newsgroup.
> 3) When you download article headers for a group, maintain a single
> composite
> list of article headers. Each entry in the list of article headers
> will contain a new 32 bit field which will have a bit set for each
> server which contains the article.
> 4) When trying to download the articles, try the valid servers in
> sequence until the article is successfully downloaded.
>
> Beyond the necessary cleanup issues (for example, when you delete
> a server), why is it more complicated than this?
Many servers don't handle retrieving bodies keyed by Message-ID;
using the group's index number for the article is more reliable.
To do this we'll have to map between message-id and
server-group-index tuples, as described in
http://mail.freesoftware.fsf.org/pipermail/pan-devel/2002-October/000220.html .
Also I think we need to tackle how to best define the new/unread/deleted
article lists in a DB setting. More background on this is in the same link.