moving last mail from nnmh to nnml fails

moving last mail from nnmh to nnml fails

[Matt Swift]

I can't move (with `B m') my last piece of mail from any nnmh group to
any nnml group.  I'm trying to migrate a *lot* of stuff from MH, and I
can't do it (without a lot of manual intervention) until I get around
this problem.

`nnmail-keep-last-article' is nil

The messages I get when trying to do this are:

    Moving to nnml+corr:TeX: (1)...
    Group Miscellaneous selected
    Couldn't Move article 1: Group Miscellaneous selected

Where "Miscellaneous" is the group I am trying to move mail from.

I can move all mail from an nnml group to an nnml group.

Before I investigate this further, do other people agree this is a
bug?  (occurs in 0.98 and also latest CVS version; I've noticed this
behavior since at least 0.96)

Re: moving last mail from nnmh to nnml fails

[Shenghuo ZHU]

>>>>> "Matt" == Matt Swift <swift@alum.mit.edu> writes:

Matt> I can't move (with `B m') my last piece of mail from any nnmh group to
Matt> any nnml group.  I'm trying to migrate a *lot* of stuff from MH, and I
Matt> can't do it (without a lot of manual intervention) until I get around
Matt> this problem.

Matt> `nnmail-keep-last-article' is nil

Matt> The messages I get when trying to do this are:

Matt>     Moving to nnml+corr:TeX: (1)...
Matt>     Group Miscellaneous selected
Matt>     Couldn't Move article 1: Group Miscellaneous selected

Matt> Where "Miscellaneous" is the group I am trying to move mail from.

Matt> I can move all mail from an nnml group to an nnml group.

Matt> Before I investigate this further, do other people agree this is
Matt> a bug?  (occurs in 0.98 and also latest CVS version; I've
Matt> noticed this behavior since at least 0.96)

(setq nnmh-allow-delete-final t)

-- 
Shenghuo ZHU

Re: moving last mail from nnmh to nnml fails

[Matt Swift]

>>"Z" == Shenghuo ZHU <zsh@cs.rochester.edu> writes:

>>>>> "Matt" == Matt Swift <swift@alum.mit.edu> writes:
    Matt> I can't move (with `B m') my last piece of mail from any
    Matt> nnmh group to any nnml group.  

    Z> (setq nnmh-allow-delete-final t)


Thanks.  I found that one quite easily, but I couldn't access this
server to say so for a while. 

This variable is not customized, not in the .info documentation, and
doesn't have a doc string.

Also, I don't understand its default setting of nil (or, indeed, the
reason the variable exists).  MH has no trouble dealing with empty
folders.  

Or, perhaps it is cleanest and clearest if the functions that check
`nnmh-allow-delete-final' should check `nnmail-keep-last-article'
instead.

Re: moving last mail from nnmh to nnml fails

[Kai Großjohann]

Matt Swift <swift@alum.mit.edu> writes:

> Also, I don't understand its default setting of nil (or, indeed, the
> reason the variable exists).  MH has no trouble dealing with empty
> folders.  

nnmh looks in the directory and uses the next free article number.
And Gnus doesn't like it at all if article numbers are reused.
Consider: you have marked as read all articles from 1 through 1000.
You delete all articles from the group.  You put a new article in the
group.  It gets number 1 (since that's the next free number).  The new
article will be marked as read!

kai
-- 
This gubblick contains many nonsklarkish English flutzpahs,
but the overall pluggandisp can be glorked from context. -- David Moser

Re: moving last mail from nnmh to nnml fails

[Matt Swift]

>>"K" == Kai Großjohann <Kai.Grossjohann@CS.Uni-Dortmund.DE> writes:

    K> Matt Swift <swift@alum.mit.edu> writes:
    >> Also, I don't understand its default setting of nil (or,
    >> indeed, the reason the variable exists).  MH has no trouble
    >> dealing with empty folders.

    K> nnmh looks in the directory and uses the next free article
    K> number.  And Gnus doesn't like it at all if article numbers are
    K> reused.  Consider: you have marked as read all articles from 1
    K> through 1000.  You delete all articles from the group.  You put
    K> a new article in the group.  It gets number 1 (since that's the
    K> next free number).  The new article will be marked as read!

I see your point.  Aren't analogous problems going to occur if you use
another MH client to manipulate a folder between Gnus sessions?  It
seems reasonable to need to avoid using another MH client during a
single Gnus session, but not reasonable to need to avoid ever using
another MH client if you want to use Gnus as an MH client on occasion.

nnmh is trying to be more than a mere MH client, by keeping track of
read articles (and possibly other state) that MH does not comprehend.
For some people, this might be helpful.  I wonder if there is room for
a flavor of nnmh that is friendlier when working with other MH
clients.

I will ponder.

Not that I've observed this problem ... yet.  Maybe there isn't one.

Re: moving last mail from nnmh to nnml fails

[David S. Goldberg]

To use another MH client and gnus/nnmh is no problem as long as you
don't ever do a folder -pack or empty the folder with the other
client.  At one point, back in the September or Red days as I recall,
there was talk, a lot of it by me, of making nnmh grok the
.mh_sequences file and use that for sharing mark info with other MH
clients.  At one point I actually had a few bits of it sort of
working, namely gnus would query the unseen sequence and mark
everything else as read.  Unfortunately, it wasn't automatic in that I
had to run a separate command in summary buffer to do it (see, I said
gnus there, not nnmh :-)

And then I switched to nnml and message-mode and I haven't so much as
looked at MH since.
-- 
Dave Goldberg
Post: The Mitre Corporation\MS B325\202 Burlington Rd.\Bedford, MA 01730
Phone: 781-271-3887
Email: dsg@mitre.org

Re: moving last mail from nnmh to nnml fails

[Matt Swift]

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

    D> To use another MH client and gnus/nnmh is no problem as long as
    D> you don't ever do a folder -pack or empty the folder with the
    D> other client.  

Thanks.

I pack all my MH folders every night in a cron job, unfortunately....

    D> And then I switched to nnml and message-mode and I haven't so
    D> much as looked at MH since.

It seems that during my migration from MH -> nnml, I could make sure
that I ignore Gnus's idea of what articles are read and unread in
nnmh groups.

There's nothing to watch out for besides the read/unread status, is
there?

Re: moving last mail from nnmh to nnml fails

[Kai Großjohann]

Matt Swift <swift@alum.mit.edu> writes:

> I pack all my MH folders every night in a cron job, unfortunately....

Why don't you turn that off?  Article numbers aren't important, for
Gnus at least.

kai
-- 
This gubblick contains many nonsklarkish English flutzpahs,
but the overall pluggandisp can be glorked from context. -- David Moser

Re: moving last mail from nnmh to nnml fails

[Matt Swift]

>>"K" == Kai Großjohann <Kai.Grossjohann@CS.Uni-Dortmund.DE> writes:

    K> Matt Swift <swift@alum.mit.edu> writes:
    >> I pack all my MH folders every night in a cron job,
    >> unfortunately....

    K> Why don't you turn that off?  Article numbers aren't important,
    K> for Gnus at least.

I can do without the packing, but I realize sorting my MH folders is
going to cause similar problems, and having my mails in chronological
order is a pretty important thing to me when reading over an email
exchange.

I'm just not going to mark any articles in nnmh groups `read' with
Gnus, and hurry up and get on with my migration so I can bag MH for
good -- i still have to set up Gnus posting styles, mail
abbreviations, mail sources, mail splitting, etc. etc.