uudecoding bug?

uudecoding bug?

[Vladimir Volovich]

Hi,

i noticed that gnus refuses to "uudecode and save" a file from
messages which contain 8-bit (russian koi8-r) subjects. it seems that
gnus looks for a sequence of articles with `similar' subjects to find
uu-parts, but 8-bit characters in subjects seem to confuse gnus, and
it does not even try to find the messages with uuencoded file.

	Best regards, -- Vladimir.

Re: uudecoding bug?

[Lars Magne Ingebrigtsen]

Vladimir Volovich <vvv@vvv.vsu.ru> writes:

> i noticed that gnus refuses to "uudecode and save" a file from
> messages which contain 8-bit (russian koi8-r) subjects.

Is this the `X U' command?

> it seems that gnus looks for a sequence of articles with `similar'
> subjects to find uu-parts, but 8-bit characters in subjects seem to
> confuse gnus, and it does not even try to find the messages with
> uuencoded file.

What does

(gnus-uu-reginize-string (gnus-summary-article-subject))

on one of these lines say?

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

Re: uudecoding bug?

[Vladimir Volovich]

"LMI" == Lars Magne Ingebrigtsen writes:

 LMI> Is this the `X U' command?

Yes.

 LMI> What does
 LMI> (gnus-uu-reginize-string (gnus-summary-article-subject))
 LMI> on one of these lines say?

Here is it (in russian):

"Шрифт[ 	]+на[ 	]+обсуждение[ 	]+(TTF)[ 	]+([0-9]+/7)"

After pressing `X U', gnus asks about the directory where to save, and
when i press enter, nothing happens (gnus writes `saved 0 files', and
cursor remains in the summary buffer).

	Best regards, -- Vladimir.

Re: uudecoding bug?

[Lars Magne Ingebrigtsen]

Vladimir Volovich <vvv@vvv.vsu.ru> writes:

>  LMI> What does
>  LMI> (gnus-uu-reginize-string (gnus-summary-article-subject))
>  LMI> on one of these lines say?
> 
> Here is it (in russian):
> 
> "Шрифт[ 	]+на[ 	]+обсуждение[ 	]+(TTF)[ 	]+([0-9]+/7)"

That looks ok.

Does 

(gnus-uu-find-articles-matching)

over one of those lines return `nil'?  If so, could you edebug-defun
that function?

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

Re: uudecoding bug?

[Vladimir Volovich]

"LMI" == Lars Magne Ingebrigtsen writes:

 LMI> Vladimir Volovich <vvv@vvv.vsu.ru> writes: What does
 LMI> (gnus-uu-reginize-string (gnus-summary-article-subject)) on one
 LMI> of these lines say?
 >>  Here is it (in russian):
 >> 
 >> "Шрифт[ ]+на[ ]+обсуждение[ ]+(TTF)[ ]+([0-9]+/7)"

 LMI> That looks ok.
 LMI> Does
 LMI> (gnus-uu-find-articles-matching)
 LMI> over one of those lines return `nil'?

Yes.

 LMI> If so, could you edebug-defun that function?

Well, i'm not so much a elisp hacker. :-) I did not find how to use
edebug-defun. I used debug-on-entry instead. Here is what happens:
gnus prepared a regexp to look for subjects, and then started to
verify matching for all subjects in the summary buffer. i.e. it
executed statements like this:

string-match("\xe48\xe60\xe58\xe64\xe62[ 	]+\xe5d\xe50[ 	]+\xe5e\xe51\xe61\xe63\xe56\xe54\xe55\xe5d\xe58\xe55[ 	]+(TTF)[ 	]+([0-9]+/7)" "Re: \xe3a\xe50\xe5f\xe58\xe62\xe55\xe5b\xe6c \xe51\xe63\xe5a\xe52\xe6b \"\xe67\".")

All of them returned nil, and no articles were found. Maybe, this is
because of multibyte/encoding/whatever?

	Best regards, -- Vladimir.

Re: uudecoding bug?

[Vladimir Volovich]

"LMI" == Lars Magne Ingebrigtsen writes:

 LMI> If so, could you edebug-defun that function?

Well, here are some more testing results. The following two
string-match calls return nil for me, but they should return t,
because these are for articles with the same thread. Is this a bug in
Emacs or in gnus?

(string-match "\xe48\xe60\xe58\xe64\xe62[ 	]+\xe5d\xe50[ 	]+\xe5e\xe51\xe61\xe63\xe56\xe54\xe55\xe5d\xe58\xe55[ 	]+(TTF)[ 	]+([0-9]+/7)" "\xe48\xe60\xe58\xe64\xe62 \xe5d\xe50 \xe5e\xe51\xe61\xe63\xe56\xe54\xe55\xe5d\xe58\xe55 (TTF) (1/7)")

(string-match "\xe48\xe60\xe58\xe64\xe62[ 	]+\xe5d\xe50[ 	]+\xe5e\xe51\xe61\xe63\xe56\xe54\xe55\xe5d\xe58\xe55[ 	]+(TTF)[ 	]+([0-9]+/7)" "\xe48\xe60\xe58\xe64\xe62 \xe5d\xe50 \xe5e\xe51\xe61\xe63\xe56\xe54\xe55\xe5d\xe58\xe55 (TTF) (4/7)")

	Best regards, -- Vladimir.

Re: uudecoding bug?

[Vladimir Volovich]

"VV" == Vladimir Volovich writes:

 VV> Well, here are some more testing results.

Just one more very strange result:

(string-match "Шрифт" "Шрифт")
nil

Looks like a bug in Emacs. ;-(

	Best regards, -- Vladimir.

Re: uudecoding bug?

[Lars Magne Ingebrigtsen]

Vladimir Volovich <vvv@vvv.vsu.ru> writes:

> Just one more very strange result:
> 
> (string-match "Шрифт" "Шрифт")
> nil
> 
> Looks like a bug in Emacs. ;-(

Yes.  You could report this to RMS just to be sure, but I think I
brought a related issue up with him, and he said it way supposed to be 
fixed in the latest pretest.  But I haven't tested it myself yet.

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