Re: Bugs in PGnus

[Brian May <bmay@csse.monash.edu.au>]

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

    Lars> If you telnet to the server and say just

    Lars> HEAD <38385A79.8410721@mcn.net>

    Lars> to it, do you get the complete header, including this Xref line? 

Yes.

head 31325
[...]
Xref: towncrier.cc.monash.edu.au gnu.emacs.gnus:31325

and

head <87u2m2482f.fsf_-_@tux.tuxoncd.de>
[...]
Xref: towncrier.cc.monash.edu.au gnu.emacs.gnus:31325

while some things are messed up on this news server, that
seems to be OK.

xover 31325
224 data follows
31325   Re: .procmailrc and gnus - sorry        Holger Nassenstein <h.nassenstein@gmx.de>       Wed,  1 Dec 1999 23:24:07 +0100 <87u2m2482f.fsf_-_@tux.tuxoncd.de>      <87puwsxalx.fsf@tux.tuxoncd.de> <vaf3dtn6592.fsf@lucy.cs.uni-dortmund.de>       1900    42      Xref: towncrier.cc.monash.edu.au gnu.emacs.gnus:31325
.

xover <87u2m2482f.fsf_-_@tux.tuxoncd.de>
224 data follows
.


(I suspect this is an illegal command).


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

    Lars> Brian May <bmay@csse.monash.edu.au> writes:

    >> 2. As I already mentioned, threading often has problems, as messages
    >> are put with the wrong thread. This seems to be because "Y c"
    >> inserts cached articles at the *current* cursor position, and not in
    >> thread order. Pushing "T t" on the broken thread fixes it.

    Lars> Does this only happen with the cached articles, or with the other
    Lars> similar commands (`j', `^', `M-^') as well?

j - yes.
M-^ - yes, same problem for both.

^ - no, I would expect articles to appear above the cursor anyway.
This doesn't mean there isn't a problem though.

All so, something I didn't expect, M-^ will fetched the article
from another newsgroup and insert into the current group. This
is probably normal, but unexpected for me.

If this is the case, the IMHO, followups to multiple
messages should be cross posted to both groups (currently
it only goes to one group).

    >> 4. I sent myself a HTML file (please ask if you want a copy) that
    >> displays OK in X-Windows, but in a text-mode console, the top
    >> of the file is missing,

    Lars> Could it be that w3 renders the message in a font that's not, er,
    Lars> visible?

I don't think so (at least not for all messages).

    >> According to weblint, the "content=" should be quoted. However,
    >> I didn't write this HTML, so it is outside my control. PGnus
    >> gets *very* confused and displays:
    >> 
    >> html;charset=iso-8859-1 http-equiv=Content-Type>

    Lars> This has been fixed in more recent versions of w3.

Yes.  I suddenly realized the version I was using is an early
beta version, supplied with xemacs (I think).

    >> Another problem with this message is that in text mode xemacs,
    >> the actual content is invisible (black on black) - probably
    >> because the HTML sets the background color, but not the
    >> foreground color - Stupid!  No wonder - it is "MSHTML" ;-).

    Lars> There are w3 variables for customizing a bunch of font
    Lars> things.  Newer version of w3 are very good at coping with
    Lars> all sorts of totally broken HTML.

There is one web page where the newest W3 version is worse then
the old - http://dictionary.com/ - none of the content is shown.

I tried to send a bug report with the automatic bug utility (which
didn't use Gnus, but its own message editor :-( ), but haven't
received a reply yet.

However, the newer version generally seems better.

Oh... on the topic of bug reports, I tried to send a bug report on
supercite, as it gets confused with some "From" addresses[1], but the bug
report bounced. Anyone know the correct address for sc bug reports?

(note[1]: is this an legal address? eg
"Brian=20A.=20May"@somwhere.fidonet.org, which my Fidonet gateway
converted (including quotes) from the Fidonet address "Brian A. May" )

    >> However, while PGnus will display an entry for the cached article, it
    >> refuses to display the contents of this article:
    >> 
    >> Symbol's value as variable is void: gnus-article-decode-hook

    Lars> That's very odd.  (Time passes.)  Aha!  It's the old
    Lars> binding-a-variable-before-its-defvar-is-loaded thing.  Fix in cvs.

Thanks.

PS. I haven't tried PGnus 0.99 yet, will do so ASAP.
-- 
Brian May <bmay@csse.monash.edu.au>