active file too far ahead

[Per Hedeland]

>Gnus uses LIST ACTIVE when starting up to find out how much new news
>has arrived.  But it will only use the answer from GROUP to actually
>increase its own highmark in .newsrc.eld.

That does seem rather stupid (though not fatal in any way of course) -
given the state of the world, I think you should report it as a bug; as
I wrote the potential for GROUP giving a lower himark than active is
certainly nothing INN-2.x-specific, but rather the norm - regardless of
whether it's "right" or not.

>I don't understand why it is bad to have GROUP return the _real_
>highmark.

I don't think it is bad at all, just noted that it may well not do that
- the reason it traditionally hasn't is presumably that implementors
have taken the wording of 977 to mean that they should try to give an
answer that is as "truthful" as possible, in particular that not just
the number of articles but also the lo- and hi-marks should correspond
to the current actual contents of the group.

I can't see any problems with changing this such that GROUP's himark
will be the one from active, clients must be able to cope with a missing
himark article anyway, as it could be canceled between GROUP and
retrieval attempt. And then at least *your* Gnus will work.:-)

--Per