From xemacs-m  Wed Jan 29 16:53:44 1997
Received: from cdc.noaa.gov (manager.Colorado.EDU [128.138.218.210])
          by xemacs.org (8.8.4/8.8.4) with SMTP
	  id QAA05889 for <xemacs-beta@xemacs.org>; Wed, 29 Jan 1997 16:53:42 -0600 (CST)
Received: from charney by cdc.noaa.gov (SMI-8.6/SMI-SVR4)
	id PAA23468; Wed, 29 Jan 1997 15:53:38 -0700
Received: by charney (SMI-8.6) id PAA20982; Wed, 29 Jan 1997 15:53:37 -0700
Sender: mdb@cdc.noaa.gov
To: Sudish Joseph <sudish@mindspring.com>
Cc: XEmacs beta-list <xemacs-beta@xemacs.org>,
        gnus-bug@ifi.uio.no (Lars Magne Ingebrigtsen)
Subject: Re: [PATCH!] Re: [rgnus-0.83] mark behavior change?
References: <vkg1zqzx9u.fsf@cdc.noaa.gov> <m2n2tyt96a.fsf@proletcult.slip.ifi.uio.no> <vkn2tvf07k.fsf@cdc.noaa.gov> <m2zpxvklkq.fsf@proletcult.slip.ifi.uio.no> <vkafpu52qn.fsf@cdc.noaa.gov> <m2k9oylqub.fsf@proletcult.slip.ifi.uio.no> <vk20b5ad9x.fsf@cdc.noaa.gov> <m2ohe9y6pg.fsf@proletcult.slip.ifi.uio.no> <vkwwswkzd7.fsf_-_@cdc.noaa.gov> <yviaenf4m9mq.fsf@atreides.mindspring.com>
Organization: CIRES, University of Colorado
X-Attribution: mb
From: Mark Borges <mdb@cdc.noaa.gov>
Date: 29 Jan 1997 15:53:36 -0700
In-Reply-To: Sudish Joseph's message of 29 Jan 1997 16:53:01 -0500
Message-ID: <vkd8uow0sv.fsf@cdc.noaa.gov>
Lines: 116
X-Mailer: Gnus v5.4.6/XEmacs 20.0

>> On 29 Jan 1997 16:53:01 -0500,
>> Sudish Joseph(SJ) wrote:
SJ> Mark Borges writes:
>> (fset 'gnus-characterp 'char-int-p)
>> Otherwise, this (near line 5417 of gnus-sum.el and perhaps elsewhere),
>> (and (gnus-characterp mark)
>> in gnus-summary-mark-article() returns nil when it should return t (as
>> it does in XEmacs-19.15).

SJ> It should return nil for non-character values of mark--all valid
SJ> article marks are characters.

I thought Lars was hinting that the problem may be that characters are
not integers in 20.0:

------- Start of forwarded message -------
From: gnus-bug@ifi.uio.no (Lars Magne Ingebrigtsen)
To: Mark Borges <mdb@cdc.noaa.gov>
Subject: Re: [rgnus-0.83] mark behavior change?
References: <vkg1zqzx9u.fsf@cdc.noaa.gov>
Date: 25 Jan 1997 10:07:25 +0100
Message-ID: <m2n2tyt96a.fsf@proletcult.slip.ifi.uio.no>

Mark Borges <mdb@cdc.noaa.gov> writes:

> I could've sworn that prior to rgnus-0.83 after I read an article in
> my nnml groups it would get marked as expirable (`E'); now it appears
> they only get marked as old (`O') and consequently never get deleted.

There have been some problems with characters in XEmacs 20 not being
integers, and this may be related.  I haven't installed XEmacs 20 here
yet, so I can't test this.

-- 
If you want an explanation on what you have just read, please 
refer to <URL:http://www.ifi.uio.no/~larsi/zombie.html>.

------- End of forwarded message -------

SJ> The test uses integerp in GNU Emacs and

actually it uses numberp in GNU Emacs:
  $ grep gnus-characterp *.el
  gnus.el:  (defalias 'gnus-characterp 'numberp)
  $

SJ> in XEmacs 19.x, but that broke the marking of articles in
                        ^^^^
what that?

SJ> auto-expirable groups under 20.x.
Yes, I agree auto-expirable groups under 20.x do not get properly
marked (see below).

SJ> Switching to char-int-p would lead to the same problems as with
SJ> intergerp, namely:
SJ> (characterp ?E) => t
SJ> (char-int-p ?E) => nil
SJ> (integerp ?E)   => nil

Yes, but

(characterp gnus-read-mark) => nil
(char-int-p gnus-read-mark) => t
(integerp gnus-read-mark)   => t

which is what the code in gnus-sum.el is testing. So maybe not even
doing the fset in gnus-xmas.el would work in 20.0?

SJ> It'd be nice to know what bug in gnus you're trying to track down.

This:

------- Start of forwarded message -------
From: gnus-bug@ifi.uio.no (Lars Magne Ingebrigtsen)
To: Mark Borges <mdb@cdc.noaa.gov>
Subject: Re: [rgnus-0.83] mark behavior change?
References: <vkg1zqzx9u.fsf@cdc.noaa.gov> <m2n2tyt96a.fsf@proletcult.slip.ifi.uio.no> <vkn2tvf07k.fsf@cdc.noaa.gov> <m2zpxvklkq.fsf@proletcult.slip.ifi.uio.no> <vkafpu52qn.fsf@cdc.noaa.gov>
Date: 27 Jan 1997 23:02:04 +0100
Message-ID: <m2k9oylqub.fsf@proletcult.slip.ifi.uio.no>

Mark Borges <mdb@cdc.noaa.gov> writes:

> I've noticed that when I read an article in XEmacs-20, the article
> gets an `R' mark. In contrast, when I read an article in XEmacs-19,
> the article gets an `E' mark.
> 
> l> The auto-expire marking is done in `gnus-summary-mark-article-as-read'
> l> in "gnus-sum.el".

Could you edebug through that function and see what happens?

-- 
If you want an explanation on what you have just read, please 
refer to <URL:http://www.ifi.uio.no/~larsi/zombie.html>.

------- End of forwarded message -------

run edebug-defun on gnus-summary-mark-article() and this line, (which
gets evalled in 19.15) 

    (setq mark gnus-expirable-mark))

never gets hit in 20.0

I have (setq gnus-auto-expirable-newsgroups "^nnml:"), and auto-expiry
does not work in 20.0. With the simple change I made, it does. Can you
explain why it works now, but didn't before?

I suppose I could be hitting a "two wrongs make a right" sort of
thing, but I'd truly like to know the proper fix.

Thanks.

-- 
  -mb-

