The status bar reported that Agent was downloading up to 14,000
headers posted since December 21 at 4:38pm. That was true of the
Earthlink server, but that wouldn't be the correct time for the
Easynews server. Earthlink had no new headers available.
Damaeus
--
Damaeus - Damon M.
That is the correct time for a Get New Headers. If you want to get old
headers you need to do a Sample Recent Headers.
--
Carroll Robbins
> That is the correct time for a Get New Headers. If you want to get old
> headers you need to do a Sample Recent Headers.
The point was that was for "new headers" from the last time I pulled
from the Earthlink news server. The time should have been different
for pulling headers from the Easynews server since I did not pull from
the Easynews server on December 21 at 4:38pm. I dunno. I'm just
reading sentencies and validating their truth. My status bar lied to
me and I was just talking about it here.
> I have been using just Earthlink's news server to access this group
> lately, but I just switched the folder's properties to also download
> from Easynews after noticing that some messages have been threaded
> under mine, but which were not replies to my posts. Thinking I might
> be missing some messages, I added Easynews to the access plan.
If this mis-threading is caused by the Agent hash collision problem,
then this will not help.
> The status bar reported that Agent was downloading up to 14,000
> headers posted since December 21 at 4:38pm. That was true of the
> Earthlink server, but that wouldn't be the correct time for the
> Easynews server. Earthlink had no new headers available.
This date is cosmetic. Don't let it worry you more than it is worth.
> Damaeus
--
Cheers,
Ralph
> This date is cosmetic. Don't let it worry you more than it is worth.
I wasn't worried about it, really, but I figured since I noticed it, I
might as well say something about it.
In particular, the previously posted "Table of collision probabilities"
is very interesting. This is supposed to be the probability of the
error occurring depending on how many messages are in a group.
Anecdotally, I would say those stats pretty much hold true for me...
FWIW, I believe the probabilities are a little higher.
The posted "Table of collision probabilities" is based on the
assumption that all hash values are equally likely.
It looks to me that for MIDs, approximately half of the hash values
will not occur in practice. The effect of this would be that the
group would only need approximately 70% of the numbers shown in the
table to achieve the corresponding probability of a hash collision.
--
Cheers,
Ralph