On Thu, Oct 02, 2003 at 04:07:05PM +0300, Daniel Vainsencher wrote:
> Seems to me we could have the same effect with less work by having a
> "time window filter" in the BFAV that shows only the last years posts.
> That would make it easy both to do the regular harvesting/reviewing, but
> also to go fishing for oldies.
> 
> What would change in essence is only that we'd be saying that the
> harvester generally look only at the stuff that either recent or
> recently touched. Which I think would reasonable...
> 
This is actually what happens right now: no fix/enh from 2002 or
earlier was harvested in the last months. I just changed my BFAV
to show items back till nov 2002, and it's (even after allready
sending lots of closed messages) a mess. There's no chance
that we will ever harvest stuff older than 2003, and currently
harvesting is slower than new submissions.

What I wanted was just to make this explicit: make it clear to
those looking at the archive, the author and all users
that this submission won't be looked at. Yes, this is not
nice, and I agree that the way it should be is that we
look at everything oldest first. But this is just imposible
without paying someone for it: Right now, some people (and
not that many, I might add) harvest stuff if they understand 
it and are sure that they want it. And this will not change in the future...

So we have two alternatives
a) non-harvested changes will be forgotten without sending
   an email to anyone, with some people hoping that they
   will be dealed with eventually (which wont happen).
b) non-harvested changes will be put in a defined state,
   with mail sent to all parties (list, author) so he might
   start to bug some people to have a look at it.

I personally think both a) and b) can be good, b) is for
the situation like we have now: a river of enh streaming 
through, with some brave souls fishing for stuff to harvest,
whereas a) is best if the set of non-harvested changes is 
small.

   Marcus


-- 
Marcus Denker marcus at ira.uka.de  -- Squeak! http://squeak.de