[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [leafnode-list] [ANNOUNCE] leafnode-2.0b5



krasel@xxxxxxxxxxxxxxxxxxxxxxxxxxxx (Cornelius Krasel) writes:

>> BTW, I have some ideas how to improve the supersedes processing.
>> Anyone interested in?
>
> Sure :-)

Currently, supersede() completely removes the superseded article
from the spool and from leafnode's memory.

Due to article propagation delays or flawy feeds, one upstream
server may present the original article after the supersede has
already been received from another server. In this case, the
original will be (re-)inserted into the spool. To prevent this,
some kind of history is necessary.

I've modified supersede() to create a new (empty) file in the
message.id/ tree if the superseded article is not found. Then a
link to the article is created in a new directory, which I called
"super.seded/". After that, the article gets removed from the
group directories, but kept in message.id/.

Up to here, this seems to work. Even pulling news from an "old"
leafnode (without Supersedes handling) using fetchnews -x does no
harm.

At this state, leafnode still allows to retrieve the superseded
message via Message-ID, even if the file is empty. To save disk
space, the article may be truncated to zero length if desired.

The super.seded/ entries should be expired somehow. Texpire will
then remove the remaining message.id entries.

Stefan


-- 
leafnode-list@xxxxxxxxxxxxxxxxxxxxxxxxxxxx -- mailing list for leafnode
To unsubscribe, send mail with "unsubscribe" in the subject to the list