Very unhappy (again). What happened? "B r" marked all unread as read.

Very unhappy (again). What happened? "B r" marked all unread as read.

[Rob Browning]

I was trying to clear up some long-term problems with my split rules,
and so after I figured out what I really wanted, I changed the rules,
and then tested "B r" on a ticked message.  This was to make sure gnus
was preserving marks across respools.  Last time (about a year ago?) I
tried it, it didn't, and I lost marks for (as I recall) over 5000
messages.  I still haven't recovered from that -- I just left that
group (my old inbox) to be dealt with later, and hoped that any things
I missed, people would just badger me about.  This time I thought I'd
be safer, and in fact, in the test this time, the tick was preserved.

So I thought it would be OK to proceed.  I marked *all* the messages
in my inbox (a few thousand) with the process mark, and then ran a "B
r".  After grinding for a while, it finished.  Along the way, it just
converted all my unread messages to read (i.e. "O").  And now they're
mixed in with all the stuff that was scheduled for deletion (hundreds
of messages) @#$%@#$%#&$%^#!

Is this intentional?  I really hope not.

<ignorable ramble>

Every time I think I've gotten conservative enough with my gnus usage,
and I try something to make my mail mess saner, it seems like gnus
just kicks me in the head again.  Last time, gnus just killed all the
marks.  I guess this is at least an improvement.

I *REALLY* wish gnus would get backends that store the marks with the
group directory -- I'll need to look in to the maildir backend maybe.
At least then there would be a sane way to protect yourself against
this kind of thing: "cp -a inbox inbox.safe.tmp", do potentially
really dangerous thing with gnus, and if failure, quit emacs, "rm -r
inbox; mv inbox.safe.tmp inbox".

In this case, I did cp -a my inbox, ahead of time, but that doesn't
really help since now it's out of sync with all the marks in the
newsrc.eld file.  And even if I'd copied that too, it would be out of
sync with whatever messages got duplicated or moved to other
directories during the re-split.  There seems to be no way, short of
copying my entire mail tree, which is hundreds of megabytes, before
any even slightly questionable action.  Trouble is that I only find
out what's questionable *after* it's too late.

A good friend of mine switched to mutt a while ago after a couple of
similar experiences to the above.  He's still not happy there, but at
least things aren't getting trashed as often.

Don't get me wrong, there are a *lot* of things I really like about
gnus, and I'd really rather work them out, but it seems like I just
*keep* getting nailed.

</ignorable ramble>

Sorry for the griping, but I'm just upset...

-- 
Rob Browning <rlb@cs.utexas.edu> PGP=E80E0D04F521A094 532B97F5D64E3930

Re: Very unhappy (again). (Apologies)

[Rob Browning]

I'd like to apologize a bit for my tone, I definitely appreciate the
work the developers do, I was just a little stunned to find myself
again falling afoul of bad gnus karma :>

Regardless, I'd still like to report that 5.8.8 appears to lose the
unread state of messages that are "B r" respooled.  I glanced at the
source, but it wasn't immediately obvious to me what the problem is.

Thanks.

-- 
Rob Browning <rlb@cs.utexas.edu> PGP=E80E0D04F521A094 532B97F5D64E3930

Re: Very unhappy (again). What happened? "B r" marked all unread as read.

[Kai Großjohann]

I agree that a way to save marks along with the backend is a good
idea.  I wonder if it is difficult to do.  Hm.

kai
-- 
Be indiscrete.  Do it continuously.

Re: Very unhappy (again). What happened? "B r" marked all unread as read.

[Paul Stevenson]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:

> I agree that a way to save marks along with the backend is a
> goodidea.  I wonder if it is difficult to do.  Hm. 

M3 T0)!!1!.  I've lost things like 'answered' marks before
which was a pain.  Can I just say to whoever wants to implement the
Great Mark Plan:

· User-defined marks would be nice
· Is it necessary to differentiate between primary and secondary
  marks? It'd at least be nice to add some functionality to secondary
  marks, like limiting to them.

-- 
.   Paul Stevenson   .   Research Fellow   .   University of Surrey     .
.     phone: +44 (0) 1483 876795     .     fax +44 (0) 1483 876781      .
.    p.stevenson@surrey.ac.uk  .  http://www.ph.surrey.ac.uk/~phs3ps    .
.                           I dream in beige                            .