[leafnode-list] Annoying bug - was:1.9.11: article numbers (for some groups) have been

[otr <o.rousseaux@xxxxxxxxxxxxxx>]

Sorry if I am wrong, but I think this bug already exists in 1.9.14.

I explain:
When "fetchnews" removes a group from "interesting.groups" directory because
no access was made during timeout_long days and the -n option was not used,
the "high water mark" (and "low water mark") are set to one (!).
Previously fetched articles (if any) however are still stored.

If I run "texpire" just after, high mark is reset correctly and all will be
OK.
But if I don't and try with a N.R. to access this group I will find it empty
and leafnode will of course re-create the group in the "interesting.groups"
directory.
If I then run "fetchnews", articles for this re-actived group are fetched and
the numbering will start from 2 (if there is no article 2 of course - the
program skips silently over already existing 'numbers').
This leads to some confusion...

I think those two lines in fetchnews.c should be suppressed in future release:
[...]
	    unlink(s);
->	    g->first = 1;
->	    g->last = 1;
	    return server ;
[...]

High water mark should never be reset and this should also solve point 1 from
the "TODO" list for fetch.
We can go up at last to number: 2^31 - 1 (=2147483647) assuming integer are 32
bits long.

Thank you for your reading.

Jeff Grossman wrote:
> 
> That is a problem that was found in 1.9.11.  It is fixed in the 1.9.12
> release.
> 
> Jeff
> ---
> Jeff Grossman (jeff@xxxxxxxxxxx)
> 
> ----- Original Message -----
> From: "Alan Mackenzie" <acm@xxxxxx>
> To: <leafnode-list@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>
> Sent: Wednesday, April 05, 2000 4:56 AM
> Subject: [leafnode-list] 1.9.11: article numbers (for some groups) have been
> reset to 2. Help!
> 
> > To access Usenet, I am running Leafnode-1.9.11 and tin-1.4.0 on a Linux
> > system. I don't run texpire (I'm working on a program to enable me to
> > remove unwanted articles regardless of their age :-), so some of my spool
> > directories have many thousand articles in them, the largest currently
> > having 14663 articles.
> >
> > On Monday (3rd April) when I started "fetchnews -vvv" as usual:
[...]
> > (iii) The articles which were fetched for each of these three newsgroups
> > were given article numbers 2, 3, 4, ...... rather than the expected
> > 15302, 15303, ....etc.;
> >
> > (iv) Articles from the other three active groups were read normally.
> >
> > I have run fetchnews several times since this happened. In the three
> > affected groups, the article numbers being given to new articles continue
> > to increment from the very low numbers (43, 44, 45,....) rather than
> > being given the correct numbers (15302, 15303, ....).
> >
> > tin doesn't mark these articles as unread, which is inconvenient.
> >
> > QUESTIONS: What is going to happen when the increasing article numbers
> > (43, 44, 45, ....) collide with existing articles? In one group, the
> > lowest article number is around 150 :-(
> >
> > Is there some quick fix I can do to set the current article number back
> > to what it should be?
> >
> > Many thanks in advance for the help!
> >
> > --
> > Alan Mackenzie (München)
> > acm@xxxxxx

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