Moving articles destroys all marks

Moving articles destroys all marks

[Toni Drabik]

Hi,

This morning I just moved several hundred articles from one `nnfolder'
group to another (which didn't exist before, and was created when
doing `B m').

After doing `F' in *Group* buffer after that, I was unpleasantly
surprised to see that all articles in the new group appeared as unread
and without any other marks that were applied to them in the old group
(i.e. ticked articles became ordinary ones, and all `A' marks were
lost, too).

Is that the intended behavior, or bug of some kind? I believe it's the
latter, because I did some test afterwards and according to them,
articles moved to existing groups keep their marks after moving.

I looked up `Mail Groups Command' node in Gnus info (which lists
article moving commands, among the others) and haven't found anything
about this issue.

-- 
Toni Drabik <tdrabik@public.srce.hr>
Warning: This article may be fatal if swallowed.

Re: Moving articles destroys all marks

[Robert Epprecht]

Toni Drabik <tdrabik@public.srce.hr> writes:

> This morning I just moved several hundred articles from one `nnfolder'
> group to another (which didn't exist before, and was created when
> doing `B m').
> 
> After doing `F' in *Group* buffer after that, I was unpleasantly
> surprised to see that all articles in the new group appeared as unread
> and without any other marks that were applied to them in the old group

I think this is because the newly created group starts on a zombie
level, which is to low for gnus two care about marks. Create the
group in advance and set level with `S l'.

Wouldn't it make sense to let newly created groups default to the
default level?

Robert Epprecht

Re: Moving articles destroys all marks

[Toni Drabik]

Robert Epprecht <epprecht@sunweb.ch> writes:

[...]

> > After doing `F' in *Group* buffer after that, I was unpleasantly
> > surprised to see that all articles in the new group appeared as
> > unread and without any other marks that were applied to them in
> > the old group
> 
> I think this is because the newly created group starts on a zombie
> level, which is to low for gnus two care about marks. Create the
> group in advance and set level with `S l'.
> 
> Wouldn't it make sense to let newly created groups default to the
> default level?

Yes, it probably would, unless there is a good reason against that.
And I don't think there is one. We'll have to wait until Lars gets
back...

-- 
Toni Drabik <tdrabik@public.srce.hr>
Warning: This article may be fatal if swallowed.

Re: Moving articles destroys all marks

[Kai Großjohann]

Toni Drabik <tdrabik@public.srce.hr> writes:

> Robert Epprecht <epprecht@sunweb.ch> writes:
> 
> > Wouldn't it make sense to let newly created groups default to the
> > default level?
> 
> Yes, it probably would, unless there is a good reason against that.

Well, uhm, err...  I think that this is because there is already a
method for determining what happens with new newsgroups:
gnus-subscribe-newsgroup-method.  I wonder whether this is heeded when
creating groups with `B m'.

But I agree with Toni that it does not make sense for Gnus to create
the new groups as zombies.  After all, they are mail groups.  I think
new mail groups should be handled differently than new news groups.

What do you think?

kai
-- 
Life is hard and then you die.

Re: Moving articles destroys all marks

[Toni Drabik]

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

> Toni Drabik <tdrabik@public.srce.hr> writes:
> 
> > Robert Epprecht <epprecht@sunweb.ch> writes:
> > 
> > > Wouldn't it make sense to let newly created groups default to the
> > > default level?
> > 
> > Yes, it probably would, unless there is a good reason against that.
> 
> Well, uhm, err... I think that this is because there is already a
> method for determining what happens with new newsgroups:
> gnus-subscribe-newsgroup-method. I wonder whether this is heeded
> when creating groups with `B m'.

I think not. Here is what `C-h v gnus-subscribe-newsgroup-method' says
for me:

     `gnus-subscribe-newsgroup-method' is a variable declared in Lisp.

     Value: gnus-subscribe-hierarchically

-- 
Toni Drabik <tdrabik@public.srce.hr>
Warning: This article may be fatal if swallowed.

Re: Moving articles destroys all marks

[François Pinard]

Toni Drabik <tdrabik@public.srce.hr> writes:

> This morning I just moved several hundred articles from one `nnfolder'
> group to another (which didn't exist before, and was created when doing
> `B m').  After doing `F' in *Group* buffer after that, I was unpleasantly
> surprised to see that all articles in the new group appeared as unread
> and without any other marks that were applied to them in the old group
> (i.e. ticked articles became ordinary ones, and all `A' marks were
> lost, too).

By all means, I would like that moved articles recover their "unread" status.
What is the purpose of moving articles that Gnus would carefully keep out
of my sight (unless I take extraordinary measures to read them)?

I would grant that Saved or Answered status be kept with the article while
they get moved.  But definitely, I would not like any read mark on them.
The status of being tickled or dormant might be preserved or lost, yet I
would vote for loosing these, given the choice.  I definitely decide afresh
for every moved article how I want it to get lost: when moving an article,
the last thing I want is to virtually loose it, as it does not get shown.

I made a few local patches to my copy of Gnus, still 0.95, trying to erase
"readness" just before moving.  It was unbearable before.  So, I receive as
very good news that Gnus does recover the unread status spontaneously, now.
I would be tempted to call a bug whenever it does otherwise...

-- 
François Pinard   http://www.iro.umontreal.ca/~pinard

Re: Moving articles destroys all marks

[David S. Goldberg]

> By all means, I would like that moved articles recover their
> "unread" status.  What is the purpose of moving articles that Gnus
> would carefully keep out of my sight (unless I take extraordinary
> measures to read them)?

As far as I'm concerned, the purpose for moving an article is to put
it in an appropriate group.  In general what happens is I've read an
article and want to move it from my total-expire'd inbox (or some
other mail-source group that isn't especially topic oriented) to an
appropriate group based on its topic or maybe correspondent.  In
general, I see no need to read it again immediately and in those cases
I want to, I tick it and expect the tick to be preserved.  It has
worked that way for quite some time now and if mark preservation
doesn't work for groups created during the move, then that is a bug.
I'd have no trouble with an option that let you clear some or all
marks after moving should that make you happy, but I wouldn't want to
lose the present functionality.
-- 
Dave Goldberg
Post: The Mitre Corporation\MS B325\202 Burlington Rd.\Bedford, MA 01730
Phone: 781-271-3887
Email: dsg@mitre.org

Re: Moving articles destroys all marks

[Kai Großjohann]

François Pinard <pinard@iro.umontreal.ca> writes:

> By all means, I would like that moved articles recover their "unread" status.
> What is the purpose of moving articles that Gnus would carefully keep out
> of my sight (unless I take extraordinary measures to read them)?

Just for the record: I am very much used to articles staying the way
they are after being moved.  I hit M-u before moving an article if I
want it to be unread after moving.  But often, I receive a message in
my default inbox, I then deal with it, and move it to another group
for archival.  (I can't use splitting for all messages; doing the
split correctly would be AI-complete.)

kai
-- 
Life is hard and then you die.

Re: Moving articles destroys all marks

[Kai Großjohann]

David S. Goldberg <dsg@mitre.org> writes:

> As far as I'm concerned, the purpose for moving an article is to put
> it in an appropriate group.  In general what happens is I've read an
> article and want to move it from my total-expire'd inbox (or some
> other mail-source group that isn't especially topic oriented) to an
> appropriate group based on its topic or maybe correspondent.

I wonder: if a message is marked as read in a total-expirable group,
then this message is expirable, right?  Now, if you move the article
to another group, should it be expirable in the target group, as well?

By the same token, moving a message marked as read from a
non-total-expirable group to a total-expirable group might mean that
we have to mark it dormant or something, right?

What do you guys think about this?

kai
-- 
Life is hard and then you die.

Re: Moving articles destroys all marks

[David S. Goldberg]

I think that would make things way too complicated.  I like the simple
approach we have now; whatever marks are associated with the article
are preserved across the move.  That those marks might be interpreted
differently due to the use of total or auto expire is something for
the user to deal with.  Since neither auto nor total expire is set as
a default, presumably a user who sets them for a given group knows
what she is doing.  I think that changing marks behind the user's back
only leads to (probably unpleasant) surprise.
-- 
Dave Goldberg
Post: The Mitre Corporation\MS B325\202 Burlington Rd.\Bedford, MA 01730
Phone: 781-271-3887
Email: dsg@mitre.org

Re: Moving articles destroys all marks

[François Pinard]

dsg@mitre.org (David S. Goldberg) writes:

> I like the simple approach we have now; whatever marks are associated
> with the article are preserved across the move.

The simplicity is always attractive, granted.

For a long while, I've been using Gnus without marking articles as "read"
automatically on first visit, preferring to have full control over when
I really want an article to disappear out of sight.  A very instructive
thread on `ding' convinced me to try otherwise, and by now, I'm glad to
have made that move.  Oh, it is still a little irritating having to `M-u'
so often, but yet, overall, there is less typing now.  The only drawback,
and it was a serious one for me, is how easily I was "loosing" articles
while moving them around (and I do move articles a lot :-).  Forget to
`M-u', and you're done: you even have to fight a bit to recover the article
immediately, if you happen to notice your mistake.  If you do not notice,
you've just lost an article for all practical purposes.  A single mistake,
you loose more than all the time you saved by the change of paradigm, at
least for that day.  Better _never_ make mistakes, but that's too much of
a pressure for the little me, I am loosing an important part of my comfort.

My need is simple: to be sure I do not loose articles while moving them
elsewhere.  In my case, the automatic reading that comes with opening an
article is plain unwelcome when the article has to be sorted elsewhere.
My feeling is that this automatic reading should be undone, automatically as
much as possible -- so I just cannot forget to do it.  Whatever scheme that
let me sort articles as I want, and is not error-prone, is acceptable to me.

My suggestion to "unread" before moving comes from that need.

-- 
François Pinard   http://www.iro.umontreal.ca/~pinard

Re: Moving articles destroys all marks

[Kai Großjohann]

François Pinard <pinard@iro.umontreal.ca> writes:

> My need is simple: to be sure I do not loose articles while moving them
> elsewhere.  In my case, the automatic reading that comes with opening an
> article is plain unwelcome when the article has to be sorted elsewhere.
> My feeling is that this automatic reading should be undone, automatically as
> much as possible -- so I just cannot forget to do it.  Whatever scheme that
> let me sort articles as I want, and is not error-prone, is acceptable to me.

I think it would be best (and sufficient) to redefine the keys to a
function which removes all marks first (gnus-summary-mark-article nil
? ) (or do you have to write `?\ ' rather than `? '?), then calls the
normal move function.

Hm.  We want the unmarking to happen for the next N articles, or the
process-marked articles.  And we don't want the process-mark to be
removed.  Hm.  How is this done?

kai
-- 
Life is hard and then you die.

Re: Moving articles destroys all marks

[David S. Goldberg]

This is untested and almost certainly needs tweaking.

(defadvice gnus-summary-move-article (around clear-marks-first activate)
  (let ((articles (gnus-summary-work-articles (ad-get-arg 0))))
    (mapcar 'gnus-summary-clear-mark-forward articles)
    ad-do-it))

should do the right thing.  Btw, I've suggested that the clearing of
the marks be done after the move.  That was based on an assumption
that the article would be marked as read due to selection before the
move.  However a look at the source shows gnus-mark-article-hook being 
let to nil before that selection so there shouldn't be a problem.
-- 
Dave Goldberg
Post: The Mitre Corporation\MS B325\202 Burlington Rd.\Bedford, MA 01730
Phone: 781-271-3887
Email: dsg@mitre.org

Re: Moving articles destroys all marks

[François Pinard]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) écrit:
> François Pinard <pinard@iro.umontreal.ca> writes:

> > My need is simple: to be sure I do not loose articles while moving
> > them elsewhere.  In my case, the automatic reading that comes with
> > opening an article is plain unwelcome when the article has to be
> > sorted elsewhere.  My feeling is that this automatic reading should be
> > undone, automatically as much as possible -- so I just cannot forget
> > to do it.  Whatever scheme that let me sort articles as I want, and
> > is not error-prone, is acceptable to me.

> I think it would be best (and sufficient) to redefine the keys to a
> function which removes all marks first [...] then calls the normal
> move function.

This is what I did, as you might suspect, for the most common cases.

> Hm.  We want the unmarking to happen for the next N articles, or the
> process-marked articles.  And we don't want the process-mark to be
> removed.  Hm.  How is this done?

This is part of my problem!  There are many cases I do not handle properly.
I just did the most obvious or irritating cases, with some sweat and pain.
Things are more nicely done, when by Lars.  He knows Gnus pretty well :-).

But I also think I'm no extraordinary user, and that my needs could be
the needs of other users just as well, and worth sharing for this reason.
Of course, there is no single solution that makes everyone happy at once
in matters like Gnus, but the current tunability of Gnus lets me believe
that many habits can co-exist.  Maybe Lars listens? :-)

-- 
François Pinard   http://www.iro.umontreal.ca/~pinard

Re: Moving articles destroys all marks

[Toni Drabik]

François Pinard <pinard@iro.umontreal.ca> writes:

> Toni Drabik <tdrabik@public.srce.hr> writes:
> 
> > This morning I just moved several hundred articles from one
> > `nnfolder' group to another (which didn't exist before, and was
> > created when doing `B m'). After doing `F' in *Group* buffer after
> > that, I was unpleasantly surprised to see that all articles in the
> > new group appeared as unread and without any other marks that were
> > applied to them in the old group (i.e. ticked articles became
> > ordinary ones, and all `A' marks were lost, too).
> 
> By all means, I would like that moved articles recover their
> "unread" status. What is the purpose of moving articles that Gnus
> would carefully keep out of my sight (unless I take extraordinary
> measures to read them)?

When I decide to move article(s) from one group to another, I'm
usually doing it because:

a) my splitting rules failed to do the right thing; or

b) I want to archive some articles and remove them from my ``regular'' 
   groups to archiving ones.

In one case or another, I don't see why moved articles should lose
their status. I think it's rather inconvenient.

However, if you (and some others) depend on this behavior, I believe
it should be somehow customizable.


-- 
Toni Drabik <tdrabik@public.srce.hr>
Warning: This article may be fatal if swallowed.

Re: Moving articles destroys all marks

[Lars Magne Ingebrigtsen]

Toni Drabik <tdrabik@public.srce.hr> writes:

> > Wouldn't it make sense to let newly created groups default to the
> > default level?
> 
> Yes, it probably would, unless there is a good reason against that.
> And I don't think there is one. We'll have to wait until Lars gets
> back...

Hey; I'm back!

The reason it is as it is today is because of some sense of
"cleanliness"; that groups shouldn't be subscribed outside of the
normal channels.

Who cares?

So I've now made the function create and subscribe to the group you
move/copy the article to, if it doesn't exist.

-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen

Re: Moving articles destroys all marks

[Hrvoje Niksic]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> So I've now made the function create and subscribe to the group you
> move/copy the article to, if it doesn't exist.

Good choice.  *We* know we need to press `F' to see the new group, but
most regular users don't.