Bug in %B summary format?

Bug in %B summary format?

[Mike Woolley]

Hi There,

I'm using the %B summary line format and occasionally, the root of
the thread is missing the "> ". I haven't noticed any particular
pattern to this, but it does happen quite regularly. The problem also
persists between leaving and re-entering the group.

Here's an example from gnu.emacs.gnus today:

    17-Mar-03 19:56   1k   [Russ Allbery        ]  Re: pipermail archives?
 .  17-Mar-03 20:50   1k   [Paul Moore          ]  Re: Displaying HTML mai
 .  17-Mar-03 20:43   1k   `-> <Paul Moore          >  
 .  17-Mar-03 20:49   3k   [Peter Davis         ]  Re: Merging many differ
 .  17-Mar-03 23:06   1k   `-> <Johan Bockgård      >  
 .+ 17-Mar-03 23:17   2k     `-> [Lars Magne Ingebrigt]  
 .  18-Mar-03 08:32   1k       `-> <Nicolas Kowalski    >  
    18-Mar-03 03:50   1k   [Jesper Harder       ]  Re: problems with posti

Removing the customisations I've made to `gnus-summary-line-format'
(but keeping the %B"), `gnus-sum-thread-tree-single-indent' and
`gnus-sum-thread-tree-single-leaf' make no difference and the problem
still occurs.

Am I missing some other setting (wouldn't be the first time!) or is
this a bug in %B?

Cheers,
Mike

Re: Bug in %B summary format?

[Lars Magne Ingebrigtsen]

Mike Woolley <mike@ariel.co.uk> writes:

> I'm using the %B summary line format and occasionally, the root of
> the thread is missing the "> ". I haven't noticed any particular
> pattern to this, but it does happen quite regularly. The problem also
> persists between leaving and re-entering the group.
>
> Here's an example from gnu.emacs.gnus today:

[...]

>  .  17-Mar-03 20:50   1k   [Paul Moore          ]  Re: Displaying HTML mai
>  .  17-Mar-03 20:43   1k   `-> <Paul Moore          >  

As indicated by the < ... > around the name, the second article here
is not the child of the previous message, but is a gathered thread
displayed after the first message.  So it really shouldn't have a
"->" thing there at all, I think.

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

Re: Bug in %B summary format?

[Mike Woolley]

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

> Mike Woolley <mike@ariel.co.uk> writes:
>
>> I'm using the %B summary line format and occasionally, the root of
>> the thread is missing the "> ".
>> [...]
>>
>> Here's an example from gnu.emacs.gnus today:
>
> [...]
>
>>  .  17-Mar-03 20:50   1k   [Paul Moore          ]  Re: Displaying
>>  .  17-Mar-03 20:43   1k   `-> <Paul Moore          >  
>
> As indicated by the < ... > around the name, the second article here
> is not the child of the previous message, but is a gathered thread
> displayed after the first message.  So it really shouldn't have a
> "->" thing there at all, I think.

Personally, I like the `-> and would like to see a > on the pseudo
root, although the inconsistency is the most annoying thing :-)

Perhaps a new configuration option is required to control how the
above case is handled? Do you think it would be straightforward to
modify the threading code to do this? I'm quite willing to have a go
at this myself (although I'm not that familiar with the innards of
Gnus).

Cheers,
Mike

Re: Bug in %B summary format?

[Lars Magne Ingebrigtsen]

Mike Woolley <mike@ariel.co.uk> writes:

> Perhaps a new configuration option is required to control how the
> above case is handled?

I think it would probably make some sense to do it the way you want
it to be, although it would then be "lying" about what the threading
is.  Or it might make even more sense to display the gathered threads
all as roots, so they all get just ">"...

> Do you think it would be straightforward to modify the threading
> code to do this?

Pretty much.  

> I'm quite willing to have a go at this myself (although I'm not that
> familiar with the innards of Gnus).

Cool.  It's done by `gnus-summary-prepare-threads'; search for "%B"
to find the right section.  It's a long a complicated function, but
it should be somewhat understandable.

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

Re: Bug in %B summary format?

[Johan Bockgård]

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

> I think it would probably make some sense to do it the way you want
> it to be, although it would then be "lying" about what the threading
> is.  Or it might make even more sense to display the gathered threads
> all as roots, so they all get just ">"...

I like the appearance of slrn's threads, where the line goes backwards
from the pseudo root.

Since I dont like the "lying" part, I use gnus-summary-make-false-root
'dummy, and my threads look like this:

                                            Re: nnimap slowness
 R.47039 31-Mar [  Krzysztof Jędrucz ] 3 |-->Re: nnimap slowness
 R.47077 31-Mar [  Lars Magne Ingebrigt ] 1 | |-->Re: nnimap slowness
 R.47059 31-Mar [  Simon Josefsson      ] 1 | `-->Re: nnimap slowness
 R.47057 31-Mar [  Simon Josefsson      ] 1 `-->Re: nnimap slowness

The appearance I would like is the above with the actual dummy line
removed (and preferably a gnus-sum-thread-tree-false-root variable to
allow me to make the false root start ",---" or something along that
line).

=>

 R.47039 31-Mar [  Krzysztof Jędrucz ] 3 ,---Re: nnimap slowness
 R.47077 31-Mar [+ Lars Magne Ingebrigt ] 1 | |-->Re: nnimap slowness
 R.47059 31-Mar [  Simon Josefsson      ] 1 | `-->Re: nnimap slowness
 R.47057 31-Mar [  Simon Josefsson      ] 1 `-->Re: nnimap slowness


/Johan

Re: Bug in %B summary format?

[Mike Woolley]

bojohan+news@dd.chalmers.se (Johan Bockgård) writes:

> The appearance I would like is the above with the actual dummy line
> removed (and preferably a gnus-sum-thread-tree-false-root variable
> to allow me to make the false root start ",---" or something along
> that line).

I like your suggestion better than mine :-) I'll see if I can
implement it that way...

Cheers,
Mike

Re: Bug in %B summary format?

[Johan Bockgård]

Mike Woolley <mike@ariel.co.uk> writes:

> bojohan+news@dd.chalmers.se (Johan Bockgård) writes:
>
>> The appearance I would like is the above with the actual dummy line
>> removed (and preferably a gnus-sum-thread-tree-false-root variable
>> to allow me to make the false root start ",---" or something along
>> that line).
>
> I like your suggestion better than mine :-) I'll see if I can
> implement it that way...

I've been thinking about this. For it to look good there should be a
way to change the look of the summary line of a root when the subtree
under it is hidden (could be handy in general, if you want something
else then ...), to prevent something like this:

     Re: Gnus "beeping" constantly
     ,---Re: nnfolder always shows no new mails...
     Re: nnfolder/save problem...

I discovered that I could make the dummy line go away already (without
hacking gnus internals) by abusing gnus-summary-dummy-line-format (I
always wondered why you should have to end *-line-format in a "\n" in
the first place).

I have
 
(setq
   gnus-summary-make-false-root   'dummy
   gnus-summary-dummy-line-format " "
   ;; note tabulator at front
   gnus-summary-line-format
   "%1=%ua%U%R%4N %d %1{%[%z%}%* %2{%(%-20,20f%)%} %1{%]%2t %B%}%0{%S%}\n"
   )

which makes the dummy line disappear, leaving me with

     Re: Gnus "beeping" constantly
     |-->Re: nnfolder always shows no new mails
     | `-->Re: nnfolder always shows no new mails
     |   `-->Re: nnfolder always shows no new mails
     `-->Re: nnfolder always shows no new mails
     Re: nnfolder/save problem
        
But as said, that looks ugly when hiding threads, and the |--> makes
the thread look like it belongs to the previous one.

I'd be cool if you found the right way to do this.

/Johan

Re: Bug in %B summary format?

[Mike Woolley]

[-- Attachment #1: Type: text/plain, Size: 659 bytes --]

Mike Woolley <mike@ariel.co.uk> writes:

> bojohan+news@dd.chalmers.se (Johan Bockgård) writes:
>
>> The appearance I would like is the above with the actual dummy line
>> removed (and preferably a gnus-sum-thread-tree-false-root variable
>> to allow me to make the false root start ",---" or something along
>> that line).
>
> I like your suggestion better than mine :-) I'll see if I can
> implement it that way...

Patch attached.

What I've done is make a small change so that false roots are now
indicated by `gnus-sum-thread-tree-false-root'. I've defaulted this to
"> " (ie the same as `gnus-sum-thread-tree-root').

Cheers,
Mike


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: gnus-sum.el.patch --]
[-- Type: text/x-patch, Size: 1059 bytes --]

--- gnus-sum.el.orig	2003-04-12 22:47:24.000000000 +0100
+++ gnus-sum.el	2003-04-16 09:35:21.000000000 +0100
@@ -4567,6 +4567,11 @@
 If nil, use subject instead."
   :type 'string
   :group 'gnus-thread)
+(defcustom gnus-sum-thread-tree-false-root "> "
+  "With %B spec, used for a false root of a thread.
+If nil, use subject instead."
+  :type 'string
+  :group 'gnus-thread)
 (defcustom gnus-sum-thread-tree-single-indent ""
   "With %B spec, used for a thread with just one message.
 If nil, use subject instead."
@@ -4837,9 +4842,12 @@
 	     (cond
 	      ((not gnus-show-threads) "")
 	      ((zerop gnus-tmp-level)
-	       (if (cdar thread)
-		   (or gnus-sum-thread-tree-root subject)
-		 (or gnus-sum-thread-tree-single-indent subject)))
+	       (cond ((cdar thread)
+		      (or gnus-sum-thread-tree-root subject))
+		     (gnus-tmp-new-adopts
+		      (or gnus-sum-thread-tree-false-root subject))
+		     (t
+		      (or gnus-sum-thread-tree-single-indent subject))))
 	      (t
 	       (concat (apply 'concat
 			      (mapcar (lambda (item)

Re: Bug in %B summary format?

[Mike Woolley]

[-- Attachment #1: Type: text/plain, Size: 747 bytes --]

bojohan+news@dd.chalmers.se (Johan Bockgård) writes:

> I've been thinking about this. For it to look good there should be a
> way to change the look of the summary line of a root when the
> subtree under it is hidden (could be handy in general, if you want
> something else then ...)

I agree, it would be good to be able to do that, but I'm not sure how
to do it at the moment though :-)

With the change I've made for the false root string, I've got a
summary line look that I quite like now (see attached screenshot -
this is an image as I've used unicode box-drawing chars). Note that
I've set the false root to a straight arrow and a normal root to a
bent arrow, which is kind of like the opposite of slrn!

Cheers,
Mike


[-- Attachment #2: ding-shot.png --]
[-- Type: image/png, Size: 7705 bytes --]

Re: Bug in %B summary format?

[Lars Magne Ingebrigtsen]

Mike Woolley <mike@ariel.co.uk> writes:

> Patch attached.
>
> What I've done is make a small change so that false roots are now
> indicated by `gnus-sum-thread-tree-false-root'. I've defaulted this to
> "> " (ie the same as `gnus-sum-thread-tree-root').

Thanks for the patch; I've applied it to Oort Gnus v0.19 (i. e., CVS).

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

Re: Bug in %B summary format?

[Mike Woolley]

[-- Attachment #1: Type: text/plain, Size: 241 bytes --]

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

> Thanks for the patch; I've applied it to Oort Gnus v0.19 (i. e.,
> CVS).

Attached is a further patch to add the new option to the manual
(forgot about that the first time).

Cheers,
Mike


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: gnus.texi.patch --]
[-- Type: text/x-patch, Size: 582 bytes --]

--- gnus.texi.orig	2003-04-13 00:12:14.000000000 +0100
+++ gnus.texi	2003-04-17 16:10:32.000000000 +0100
@@ -4513,6 +4513,11 @@
 Used for the root of a thread.  If @code{nil}, use subject
 instead. The default is @samp{> }.
 
+@item gnus-sum-thread-tree-false-root
+@vindex gnus-sum-thread-tree-false-root
+Used for the root of a thread if it is a false root.  If @code{nil},
+use subject instead. The default is @samp{> }.
+
 @item gnus-sum-thread-tree-single-indent
 @vindex gnus-sum-thread-tree-single-indent
 Used for a thread with just one message.  If @code{nil}, use subject

Re: Bug in %B summary format?

[Lars Magne Ingebrigtsen]

Mike Woolley <mike@ariel.co.uk> writes:

> Attached is a further patch to add the new option to the manual
> (forgot about that the first time).

Thanks for the patch; I've applied it to Oort Gnus v0.19 (i. e., CVS).

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