Labels/Annotations per message

Labels/Annotations per message

[Pavel Janík ml.]

Hi,

I was thinking about the feature I wanted to have in Gnus. Very often, I need
to comment (ie. associate the annotation) with my messages. Like "I should
answer this", "This is worthy reading" and similar. I used to solve this via
creating additional nnfolder groups like AnswerIt, ReadIt etc. But this is not
universal. What if I'd like to answer/see only the messages from linux-kernel
or read only interesting things from Gnus mailing list? I do not think that
current Gnus can ease the work here. I remember that RMAIL has something like
this. So I looked around in ding archive and found the discussion between Lars
and Kai about labels. Some part of this is also in todo file:

,---- From Newest Features 
|    * LMI> Well, nnbabyl could alter the group info to heed labels like
|      LMI> answered and read, I guess.
| 
|      It could also keep them updated (the same for the Status: header of
|      unix mbox files).
|      ...
`----

I think that this will be worthy to have.

As I use nnfolder to store all my mail, I use X-Annotations header to do
that. My gnus-summary-line-format's value is 
"%U%R%z%I%(%[%4L: %-20,20f%]%) %uX%s\n"

So I prepend the result of the function gnus-user-format-function-X before the
actual subject of the message. The actual function looks like (sorry for my
elisp ;-):

(defun gnus-user-format-function-X (header)
  "My user-defined function for annotations"
  (let ((my-extra-headers (mail-header-extra gnus-tmp-header)))
    (while (and my-extra-headers (not (eq (car (car my-extra-headers)) 'X-Annotations)))
      (setq my-extra-headers (cdr my-extra-headers)))
    (setq my-extra-headers (cdr (car my-extra-headers)))
    (if my-extra-headers
	(concat "{" my-extra-headers "} ")
      "")))

And I have X-Annotations in *-extra-headers:

(setq gnus-extra-headers '(To Cc X-Annotations))
(setq nnmail-extra-headers gnus-extra-headers)

So when I receive a message with X-Annotations or I add this header via `e' to
the message in the summary buffer, it is displayed with that annotation
prepended before subject. This is sufficient for me, cause I use nnfolder
only. But this is not generic. You can not annotate nntp articles this way
etc.

So I also thought about the Real Solution (tm). We should probably invent
another type of mark, which is associated with annotations. It can be stored
in .newsrc.eld or somewhere else?

gnus-newsrc-alist in .newsrc.eld looks like this now:

("Gnus" 3 ((1 . 1748)) ((dormant 1015 (1153 . 1154)) (expire (1735 . 1748)) (reply 1466 1557 1637) (tick 1734)))

What about inventing new mark annotate like this:

("Gnus" 3 ((1 . 1748)) ((annotate (1 "This is the first message in the buffer.") (1748 "I just received this message."))))

The first message is marked as such and the last one too. If this is
implemented, you can easily limit the summary buffer to the regexp of these
annotations and similar.

What do you think about it?

I think it is worthy to discuss it here and when we will have the
specification, someone with sufficient elisp knowledge can implement it :-)
-- 
Pavel Janík ml.
Pavel.Janik@inet.cz

Re: Labels/Annotations per message

[Lars Magne Ingebrigtsen]

[-- Attachment #1: Type: text/plain, Size: 884 bytes --]

Pavel.Janik@inet.cz (Pavel Janík ml.) writes:

> So I also thought about the Real Solution (tm). We should probably invent
> another type of mark, which is associated with annotations. It can be stored
> in .newsrc.eld or somewhere else?
> 
> gnus-newsrc-alist in .newsrc.eld looks like this now:
> 
> ("Gnus" 3 ((1 . 1748)) ((dormant 1015 (1153 . 1154)) (expire (1735 . 1748)) (reply 1466 1557 1637) (tick 1734)))
> 
> What about inventing new mark annotate like this:
> 
> ("Gnus" 3 ((1 . 1748)) ((annotate (1 "This is the first message in the buffer.") (1748 "I just received this message."))))
> 
> The first message is marked as such and the last one too. If this is
> implemented, you can easily limit the summary buffer to the regexp of these
> annotations and similar.
> 
> What do you think about it?

This would be one way of doing it, but it doesn't seem to go far
enough.

[-- Attachment #2: Type: text/plain, Size: 988 bytes --]


I mean, I don't edit or annotate mail messages at all, but I know some
people who do that extensively.  (Hi, François!)  Wouldn't it be
really nice if you could edit/annotate any text in any article buffer?
You read an article in some newsgroup, and there's some useful stuff
in there, so you start editing it and inserting stuff here and there
in the article.  But you want to know which parts you wrote and which
parts were in the original article, no?

And it just now occurred to me that RCS would perhaps be the right
thing here.  The Meta-Edit command would pull the article down into
the cache, and when you edit it, you just commit your edits to the
cached version of the article.  When reading the article, Gnus would
check the latest version out, but have a special viewer that would
colorize the contributions from different people differently.  And,
and...

Er.

Anyway.

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

Re: Labels/Annotations per message

[Karl Eichwalder]

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

> And it just now occurred to me that RCS would perhaps be the right
> thing here.  The Meta-Edit command would pull the article down into
> the cache, and when you edit it, you just commit your edits to the
> cached version of the article.  When reading the article, Gnus would
> check the latest version out, but have a special viewer that would
> colorize the contributions from different people differently.  And,
> and...

Sounds good.  ... and it should be possible to copy or move or "anchor"
it to other groups.

-- 
work : ke@suse.de                          |          ------    ,__o
     : http://www.suse.de/~ke/             |         ------   _-\_<,
home : keichwa@gmx.net                     |        ------   (*)/'(*)

Re: Labels/Annotations per message

[François Pinard]

[Karl Eichwalder]

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

> > And it just now occurred to me that RCS would perhaps be the right
> > thing here.  The Meta-Edit command would pull the article down into
> > the cache, and when you edit it, you just commit your edits to the
> > cached version of the article.  When reading the article, Gnus would
> > check the latest version out, but have a special viewer that would
> > colorize the contributions from different people differently.  And,
> > and...

> Sounds good.  ... and it should be possible to copy or move or "anchor"
> it to other groups.

When `B m'ing or `B c'ing articles, annotations should move or be duplicated
accordingly.  The sad thing is that it will not work when moving articles
into, and later out from, Babyl files...

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

Re: Labels/Annotations per message

[Kai Großjohann]

On 14 Aug 2000, François Pinard wrote:

> When `B m'ing or `B c'ing articles, annotations should move or be
> duplicated accordingly.  The sad thing is that it will not work when
> moving articles into, and later out from, Babyl files...

Why not?  I thought one of the niceties of the babyl format was that
it allowed adding extra information to messages?

As a last resort, the annotations could be put in X-Gnus-* headers.

kai
-- 
I like BOTH kinds of music.

Re: Labels/Annotations per message

[Kai Großjohann]

On 13 Aug 2000, Lars Magne Ingebrigtsen wrote:

> And it just now occurred to me that RCS would perhaps be the right
> thing here.  The Meta-Edit command would pull the article down into
> the cache, and when you edit it, you just commit your edits to the
> cached version of the article.  When reading the article, Gnus would
> check the latest version out, but have a special viewer that would
> colorize the contributions from different people differently.  And,
> and...

I think you were trying to say `cvs annotate'.  No?

Sounds cool.

I kind of feel that we need both longer annotations and label-like
thingies.  The labels could go in the .overview file, so that people
can limit the summary buffer to all messages labeled todo.

kai
-- 
I like BOTH kinds of music.

Re: Labels/Annotations per message

[Lars Magne Ingebrigtsen]

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

> I think you were trying to say `cvs annotate'.  No?

I may have been, but the manual page for my cvs doesn't mention
"annotate", so I don't know.  :-)

> I kind of feel that we need both longer annotations and label-like
> thingies.  The labels could go in the .overview file, so that people
> can limit the summary buffer to all messages labeled todo.

Yup.  Labels and annotations are probably separate issues.

(But now the Marguerite Duras smash hit _India Song_ is starting on
the telly, so I'll have to watch that film again.)

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

Re: Labels/Annotations per message

[Kai Großjohann]

On 13 Aug 2000, Lars Magne Ingebrigtsen wrote:

> Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:
> 
>> I think you were trying to say `cvs annotate'.  No?
> 
> I may have been, but the manual page for my cvs doesn't mention
> "annotate", so I don't know.  :-)

`cvs annotate' prints the file, but prepends a version number to each
line.  This number says which version the line comes from.

There is a tool for colorizing the output -- red lines have recently
been changed, blue ones are old.  Or something.  `C-x v g' I think.

kai
-- 
I like BOTH kinds of music.