More fun with nnweb

More fun with nnweb

[David Z. Maze]

So I mostly have my Gnus/Debian BTS nnweb plugin working now.  So
people can follow along, it's at
http://web.mit.edu/dmaze/elisp/debbugs.el; I load it in my .gnus file.

First off, success: if I want to see a list of bugs I have open, I can 
use 'G w debbugs dmaze@debian.org'.  I'll get a listing of bugs in the 
summary buffer.  Each bug is presented as a multipart/mixed message,
and each message has a text/plain or text/enriched tag and the body,
either as message/external-body or message/rfc822.  All good.

Now, let's leave this summary buffer ('q') and try again.  From the
group buffer, I type 'G w debbugs dmaze@debian.org', and get 'No
matching articles' back from nnweb.  That's no good.  Playing around,
it seems like this is because each URL is stuffed into nnweb-hashtb,
and this isn't cleaned out before 'G w' calls.  (At least, that's what 
it seems like; if I don't play with nnweb-get-hashtb, it works fine.)

The other problem is if I try to create a persistent group (C-u G w
...).  This works fine, but each time I check for new articles ('g')
it returns duplicates of every article.  (That is, the first time, I
have 19 articles, the second, 19 new out of 38, the third, 19 out of
57, etc.)

This seems like I'm doing something slightly fundamentally wrong.
Does anybody with more nnweb clue than I have any ideas?  Thanks...

(Usage: you can feed the "search engine" anything you'd normally
concatenate to http://bugs.debian.org/.  This means you can feed it a
maintainer name ("dmaze@debian.org"), an individual package name
("xed"), or a submitter name ("from:dmaze@mit.edu").  It won't croak
badly if you feed it a bug number, but it will fail to return any bugs 
at all.)

-- 
David Maze             dmaze@mit.edu          http://www.mit.edu/~dmaze/
"Theoretical politics is interesting.  Politicking should be illegal."
	-- Abra Mitchell

Re: More fun with nnweb

[Andreas Fuchs]

[-- Attachment #1: Type: text/plain, Size: 738 bytes --]

On 2001-05-30, David Z. Maze <dmaze@MIT.EDU> wrote:
> So I mostly have my Gnus/Debian BTS nnweb plugin working now.  So
> people can follow along, it's at
> http://web.mit.edu/dmaze/elisp/debbugs.el; I load it in my .gnus file.

By the mad Goddess! This is Really Nice Stuff!

Could we arrange ourselves that yours and mine
(http://asf.void.at/emacs/gnus-BTS.el) work together? Maybe we could
even merge my solution (a rather non-gnusish one) into your debbugs
implementation?

I think that you should also mail that to Roland Mas
<lolando@debian.org>. He maintains the debbugs-el package, where my BTS
stuff is in, as you might already know.

Thanks,
-- 
Andreas Fuchs, <asf@acm.org>, <d96001@htlwrn.ac.at>, antifuchs

[-- Attachment #2: Type: application/pgp-signature, Size: 231 bytes --]

Re: More fun with nnweb

[Roland Mas]

Andreas Fuchs (2001-05-31 10:45:28 +0200) :

> I think that you should also mail that to Roland Mas
> <lolando@debian.org>. He maintains the debbugs-el package, where my
> BTS stuff is in, as you might already know.

There's no need for that.  I already ticked both articles, and I will
most certainly see what I can do about it next week.  Feel free to
provide additional info, or announce that you merged your two
packages, or whatever :-)

Roland.
-- 
Roland Mas

Sauvez les castors, imprimez en recto-verso.

Re: More fun with nnweb

[Kai Großjohann]

Sounds like great stuff (haven't had time to look at it, yet).  Now I
wish there was an interface to the reportbug program ;-)

Btw, is there a way to install the Debian BTS locally for our own
problem reports and todo items?  We are currently running Gnats but
its Web interface really sucks (it truncates lines of text entered by
the user).  Using the Debian BTS software and nndebbugs.el would
provide me with a nifty Gnus frontend, and the Emacs haters can use
the Web interface or whatever.

kai
-- 
~/.signature: No such file or directory

Re: More fun with nnweb

[Andreas Fuchs]

[-- Attachment #1: Type: text/plain, Size: 1812 bytes --]

Today, Kai Großjohann <Kai.Grossjohann@CS.Uni-Dortmund.DE> wrote:
> Sounds like great stuff (haven't had time to look at it, yet).  Now I
> wish there was an interface to the reportbug program ;-)

Err, there is one already.  Get unstable's (yeah, I know one shouldn't
recommend unstable, but...) debbugs-el package, or check out
<URL:http://asf.void.at/emacs.html> (which has a link to the original
debian-bugs.el and a patch to it), but I think you should really look
into Roland's package, because it includes even more patches to some
little bugs in debian-bugs.el itself (-:

> Btw, is there a way to install the Debian BTS locally for our own
> problem reports and todo items?  

Package: debbugs
Priority: optional
Section: misc
Installed-Size: 274
Maintainer: Darren Benham <gecko@debian.org>
Architecture: all
Version: 2.3-3
Depends: perl5 | perl, mail-transport-agent, mailtools, ed
Recommends: httpd, links | lynx
Filename: dists/woody/main/binary-all/misc/debbugs_2.3-3.deb
Size: 70212
MD5sum: ad797622b0535d318afa98a8b204f930
Description: The bug tracking system based on the active Debian BTS
 Debian has a bug tracking system which files details of bugs reported by
 users and developers. Each bug is given a number, and is kept on file until
 it is marked as having been dealt with. The system is mainly controlled by
 e-mail, but the bug reports can be viewed using WWW.
 .
 This version is fully functional, but it does not autoconfig, see
 /usr/share/doc/debbugs/README.Debian after installation.
 .
 Note: there might be some appearance issues where the a display variable
 doesn't expand as expected.

So, yes, there is. I believe there must be a version in stable, too.

regards,
-- 
Andreas Fuchs, <asf@acm.org>, <d96001@htlwrn.ac.at>, antifuchs

[-- Attachment #2: Type: application/pgp-signature, Size: 231 bytes --]

Re: More fun with nnweb

[Toby Speight]

0> In article <vaf1yp5zvlk.fsf@lucy.cs.uni-dortmund.de>,
0> Kai Großjohann <URL:mailto:Kai.Grossjohann@CS.Uni-Dortmund.DE> ("Kai") wrote:

Kai> Sounds like great stuff (haven't had time to look at it, yet).
Kai> Now I wish there was an interface to the reportbug program

You mean, like debbugs-el?

Re: More fun with nnweb

[Kai Großjohann]

On 04 Jun 2001, Toby Speight wrote:

> Kai> Sounds like great stuff (haven't had time to look at it, yet).
> Kai> Now I wish there was an interface to the reportbug program
> 
> You mean, like debbugs-el?

*blush*

kai
-- 
~/.signature: No such file or directory

Re: More fun with nnweb

[Lars Magne Ingebrigtsen]

"David Z. Maze" <dmaze@MIT.EDU> writes:

> Now, let's leave this summary buffer ('q') and try again.  From the
> group buffer, I type 'G w debbugs dmaze@debian.org', and get 'No
> matching articles' back from nnweb.  That's no good.  Playing around,
> it seems like this is because each URL is stuffed into nnweb-hashtb,
> and this isn't cleaned out before 'G w' calls.  (At least, that's what 
> it seems like; if I don't play with nnweb-get-hashtb, it works fine.)

Hmm....  Sounds like a bug.  :-)

> The other problem is if I try to create a persistent group (C-u G w
> ...).  This works fine, but each time I check for new articles ('g')
> it returns duplicates of every article.  (That is, the first time, I
> have 19 articles, the second, 19 new out of 38, the third, 19 out of
> 57, etc.)

Ah, yes, I've seen that before, but I thought I had fixed that.  The
problem was that nnweb wasn't able to create a consistent mapping
between articles and Gnus article numbers.  Do the debbugs provide
something that identifies the articles uniquely?

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

Re: More fun with nnweb

[David Z. Maze]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
LMI> "David Z. Maze" <dmaze@MIT.EDU> writes:
 DZM> The other problem is if I try to create a persistent group (C-u G w
 DZM> ...).  This works fine, but each time I check for new articles ('g')
 DZM> it returns duplicates of every article.  (That is, the first time, I
 DZM> have 19 articles, the second, 19 new out of 38, the third, 19 out of
 DZM> 57, etc.)
LMI> 
LMI> Ah, yes, I've seen that before, but I thought I had fixed that.  The
LMI> problem was that nnweb wasn't able to create a consistent mapping
LMI> between articles and Gnus article numbers.  Do the debbugs provide
LMI> something that identifies the articles uniquely?

The bug number is unique, and I should correctly parse it out of the
initial list-of-bugs page.  This seems suggest that I would want to
use it instead of (cdr active) as the number parameter to
make-full-mail-header, yes?  Is there better documentation than C-h f
for this function and how it's relevant to things?

(It looks like the way I do things now is to parse out the URL and bug 
name from the list-of-bugs page, and then, if the URL isn't already in 
nnweb-hashtb, I do
	    (push
	     (list
	      (incf (cdr active))
	      (make-full-mail-header
	       (cdr active) subject from nil nil nil 0 0 url))
	     map)
This looks like it should guarantee that each "article" has a unique
number, but if the same thing is encountered more than once with
different URLs, it winds up in the list multiple times.)

-- 
David Maze             dmaze@mit.edu          http://www.mit.edu/~dmaze/
"Theoretical politics is interesting.  Politicking should be illegal."
	-- Abra Mitchell

Re: More fun with nnweb

[Lars Magne Ingebrigtsen]

"David Z. Maze" <dmaze@MIT.EDU> writes:

> The bug number is unique, and I should correctly parse it out of the
> initial list-of-bugs page.  This seems suggest that I would want to
> use it instead of (cdr active) as the number parameter to
> make-full-mail-header, yes?

Yup.

> Is there better documentation than C-h f for this function and how
> it's relevant to things?

Nope.  There's not much documentation for nnweb.

> (It looks like the way I do things now is to parse out the URL and bug 
> name from the list-of-bugs page, and then, if the URL isn't already in 
> nnweb-hashtb, I do
> 	    (push
> 	     (list
> 	      (incf (cdr active))
> 	      (make-full-mail-header
> 	       (cdr active) subject from nil nil nil 0 0 url))
> 	     map)
> This looks like it should guarantee that each "article" has a unique
> number, but if the same thing is encountered more than once with
> different URLs, it winds up in the list multiple times.)

Yes -- you could index the table based on the bug number instead of
the URL, I think.  (It's been a while since I worked with nnweb --
ShengHuo probably has more recent experiences with the internals...) 

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