2:07 p.m.
Recently following a Debian update I found I was unable to reply to
articles in Gnus; I got an error along the lines of:
Args out of range: " .w_()'\"$\\/<>＠!|", 64
It turns out that this message comes from derived-mode, more
specifically from derived-mode-merge-syntax-tables. That function
does something like:
(when (eq ?＠ (char-syntax-from-code
(get-range-char-table key new ?＠)))
...)
The obvious intention is: "get key from range-table, or ?＠ if not
found". But char-syntax-from-code can't cope with ?＠, it expects a
small number that is a valid offset in the string. When it receives
?＠ (or 64), it signals "Args out of range".
In case it matters, the KEY for which the error occurs is the symbol
`latin-iso8859-13'. (Perhaps the Debian upgrade introduced this
charset to my XEmacs?)
Inspection revealed that ＠ happens to be at position 13 of the string
in question. Following that lead, I made this modification:
(when (eq ?＠ (char-syntax-from-code
(get-range-char-table key new 13)))
...)
which removes the error for me. But this is clearly not the right
fix because it relies on a cryptic offset into an unrelated string.
It should probably be modified to this:
(when (let ((tmpcode (get-range-char-table key new 'multi)))
(or (eq tmpcode 'multi)
(eq (char-syntax-from-code tmpcode) ?＠)))
...)
or to equivalent but more elegant code that I currently don't seem to
be able to write.
Someone with a better understanding of the code in question and of
syntax tables in general might come up with a better fix.
10:15 p.m.
* Hrvoje Niksic <hniksic(a)xemacs.org> writes:
Recently following a Debian update I found I was unable to reply to
articles in Gnus
Can you tell me which version including XEmacs package version?
(package-get-info 'gnus 'version)
is an easy way to get the XE package version
--
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
| I am Dyslexic of Borg. |
| Fusistance is retile. Your arse will be laminated. |
|------------------------------------<steve(a)sxemacs.org>---|
12:31 a.m.
Steve Youngs <steve(a)sxemacs.org> writes:
* Hrvoje Niksic <hniksic(a)xemacs.org> writes:
> Recently following a Debian update I found I was unable to reply to
> articles in Gnus
Can you tell me which version including XEmacs package version?
(package-get-info 'gnus 'version)
is an easy way to get the XE package version
XEmacs is 21.4.17, and package-get-info asks me if I want to create
some sort of index in ~/.xemacs (I don't). gnus-version indicates
version 5.10.6.
1:34 a.m.
* Hrvoje Niksic <hniksic(a)xemacs.org> writes:
Steve Youngs <steve(a)sxemacs.org> writes:
> * Hrvoje Niksic <hniksic(a)xemacs.org> writes:
>
> > Recently following a Debian update I found I was unable to reply to
> > articles in Gnus
>
> Can you tell me which version including XEmacs package version?
gnus-version indicates version 5.10.6.
Try the Gnus package in Pre-Release (XE package version 1.84).
Be aware though that 1.84 has some problems of its own (the summary
buffer menubar is broken). But I'd like to know if it fixes your
problem with replies.
Thanks.
--
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
| I am Dyslexic of Borg. |
| Fusistance is retile. Your arse will be laminated. |
|------------------------------------<steve(a)sxemacs.org>---|
9:29 p.m.
Steve Youngs <steve(a)sxemacs.org> writes:
* Hrvoje Niksic <hniksic(a)xemacs.org> writes:
> Steve Youngs <steve(a)sxemacs.org> writes:
>> * Hrvoje Niksic <hniksic(a)xemacs.org> writes:
>>
>> > Recently following a Debian update I found I was unable to reply to
>> > articles in Gnus
>>
>> Can you tell me which version including XEmacs package version?
> gnus-version indicates version 5.10.6.
Try the Gnus package in Pre-Release (XE package version 1.84).
I was fairly certain that, in my original message, I demonstrated an
error in XEmacs's implementation of derived.el that should be fixed
regardless of what Gnus is doing. In other words, using a different
version of Gnus might cause the problem to not surface, but it doesn't
mean that it shouldn't be fixed. (A simple way of fixing it is
presented in the message.)
Am I mistaken here?
10:31 p.m.
Hrvoje Niksic <hniksic(a)xemacs.org> writes:
Steve Youngs <steve(a)sxemacs.org> writes:
> * Hrvoje Niksic <hniksic(a)xemacs.org> writes:
>
> > Steve Youngs <steve(a)sxemacs.org> writes:
> >> * Hrvoje Niksic <hniksic(a)xemacs.org> writes:
> >>
> >> > Recently following a Debian update I found I was unable to reply to
> >> > articles in Gnus
> >>
> >> Can you tell me which version including XEmacs package version?
>
> > gnus-version indicates version 5.10.6.
>
> Try the Gnus package in Pre-Release (XE package version 1.84).
I was fairly certain that, in my original message, I demonstrated an
error in XEmacs's implementation of derived.el that should be fixed
regardless of what Gnus is doing. In other words, using a different
version of Gnus might cause the problem to not surface, but it doesn't
mean that it shouldn't be fixed. (A simple way of fixing it is
presented in the message.)
Am I mistaken here?
Hi Hrvoje,
I don't think you're mistaken.
Jerry has touched this code last, but I think he is relocating inside
the real world at this time.
Let's see if he's listening.
--
Adrian Aichner
mailto:adrian＠xemacs.org
http://www.xemacs.org/
2:41 a.m.
Adrian Aichner <adrian(a)xemacs.org> wrote:
Hrvoje Niksic <hniksic(a)xemacs.org> writes:
[snip]
> I was fairly certain that, in my original message, I demonstrated
an
> error in XEmacs's implementation of derived.el that should be fixed
> regardless of what Gnus is doing. In other words, using a different
> version of Gnus might cause the problem to not surface, but it doesn't
> mean that it shouldn't be fixed. (A simple way of fixing it is
> presented in the message.)
>
> Am I mistaken here?
Hi Hrvoje,
I don't think you're mistaken.
Jerry has touched this code last, but I think he is relocating inside
the real world at this time.
Let's see if he's listening.
Listening, yes, but unable to do much more than that right now. While I
touched that code last, CVS shows that this is the change that
introduced the code Hrvoje is talking about:
2003-07-18 Stephen J. Turnbull <stephen(a)xemacs.org>
* derived.el (derived-mode-merge-syntax-tables): Reverse map
logic. Fixes "do-nothing" bug.
See http://list-archive.xemacs.org/xemacs-patches/200307/msg00039.html
for the patch, and the discussion starting here:
http://list-archive.xemacs.org/xemacs-beta/200307/msg00153.html
for the rationale behind the patch. I never really understood either
the original code or the patch. Stephen, can you comment on Hrvoje's
observation?
--
Jerry James
http://www.ittc.ku.edu/~james/
5:59 a.m.
>>>> "Jerry" == Jerry James
<james(a)xemacs.org> writes:
Jerry> Stephen, can you comment on Hrvoje's observation?
Not for some days, maybe weeks. I remember that change vaguely; I
spent quite some time convincing myself that it was worth fixing the
"do nothing" bug (although I don't recall the nature of the bug
precisely).
What really needs to be done in 21.5 is to support syntax table
inheritence. I don't know if that's appropriate for 21.4 though.
--
School of Systems and Information Engineering http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
7:27 p.m.
"Stephen J. Turnbull" <stephen(a)xemacs.org> writes:
>>>>> "Jerry" == Jerry James
<james(a)xemacs.org> writes:
Jerry> Stephen, can you comment on Hrvoje's observation?
Not for some days, maybe weeks.
Then could we at least fix the bug by applying the "obvious" fix I
posted? Am I the only one for whom Gnus breaks here?
8:59 p.m.
Hrvoje Niksic <hniksic(a)xemacs.org> writes:
"Stephen J. Turnbull" <stephen(a)xemacs.org> writes:
>>>>>> "Jerry" == Jerry James <james(a)xemacs.org>
writes:
>
> Jerry> Stephen, can you comment on Hrvoje's observation?
>
> Not for some days, maybe weeks.
Then could we at least fix the bug by applying the "obvious" fix I
posted? Am I the only one for whom Gnus breaks here?
This is related to use of latin-iso8859-13, correct?
As a general question, are you still set up to commit to the XEmacs
repository?
Best regards,
Adrian
--
Adrian Aichner
mailto:adrian＠xemacs.org
http://www.xemacs.org/
12:10 a.m.
Adrian Aichner <adrian(a)xemacs.org> writes:
Hrvoje Niksic <hniksic(a)xemacs.org> writes:
> "Stephen J. Turnbull" <stephen(a)xemacs.org> writes:
>
>>>>>>> "Jerry" == Jerry James <james(a)xemacs.org>
writes:
>>
>> Jerry> Stephen, can you comment on Hrvoje's observation?
>>
>> Not for some days, maybe weeks.
>
> Then could we at least fix the bug by applying the "obvious" fix I
> posted? Am I the only one for whom Gnus breaks here?
This is related to use of latin-iso8859-13, correct?
It could be. But where is it coming from and how come I haven't run
into this before?
As a general question, are you still set up to commit to the XEmacs
repository?
Unfortunately, no.
11 a.m.
>>>> "Hrvoje" == Hrvoje Niksic
<hniksic(a)xemacs.org> writes:
> This is related to use of latin-iso8859-13, correct?
Hrvoje> It could be. But where is it coming from and how come I
Hrvoje> haven't run into this before?
If you are up-to-date with packages including new releases, my guess
is that this has to do with Aidan's fiddling with latin-unity. It
used to be that you had to enable ISO 8859/13 explicitly, but Aidan
decided that it was sufficiently important that German users have
access to DOUBLE LOW NINE QUOTATION MARK and enabled the support
unconditionally at least in 21.5. I suspect that these newly enabled
coding systems and charsets are not properly supported in the syntax
tables, etc. (I know I didn't do any of that---because I didn't, and
don't, know enough about it---and that's why I didn't enable them.)
> As a general question, are you still set up to commit to the
> XEmacs repository?
Hrvoje> Unfortunately, no.
That's a shame. However, I'm sure any of us will do what we can to
help out. It's very good just to "hear" your "voice". :-)
--
School of Systems and Information Engineering http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
12:30 p.m.
"Stephen J. Turnbull" <stephen(a)xemacs.org> writes:
>>>>> "Hrvoje" == Hrvoje Niksic
<hniksic(a)xemacs.org> writes:
>> This is related to use of latin-iso8859-13, correct?
Hrvoje> It could be. But where is it coming from and how come I
Hrvoje> haven't run into this before?
If you are up-to-date with packages including new releases, my guess
is that this has to do with Aidan's fiddling with latin-unity.
Do those changes make it into Debian packages? If so, then it's
almost certainly the case. Still, the code I've seen does have a bug.
It used to be that you had to enable ISO 8859/13 explicitly, but
Aidan decided that it was sufficiently important that German users
have access to DOUBLE LOW NINE QUOTATION MARK and enabled the
support unconditionally at least in 21.5.
Note that I'm seeing the breakage in 21.4. I also don't know what ISO
8859-13 is, but I'll look into it.
3 p.m.
>>>> "Hrvoje" == Hrvoje Niksic
<hniksic(a)xemacs.org> writes:
Hrvoje> Do those changes make it into Debian packages? If so,
Hrvoje> then it's almost certainly the case.
They do make it into the Debian packages, but how much lag, I don't
know. Not much if you're using sid, I'm not sure how fast woody and
sarge get updated.
Hrvoje> Still, the code I've seen does have a bug.
Yes. I think turning the ?＠ into a string and wrapping it in a call
to `syntax-string-to-code' should do the trick for the bug you've
identified. However, a fair amount of synching to GNU is still to be
done, and I'm sure there are plenty of bugs lurking in the syntax
code.
> It used to be that you had to enable ISO 8859/13 explicitly,
> but Aidan decided that it was sufficiently important that
> German users have access to DOUBLE LOW NINE QUOTATION MARK and
> enabled the support unconditionally at least in 21.5.
Hrvoje> Note that I'm seeing the breakage in 21.4.
Yes, however Aidan has been pretty aggressive about the Mule-related
packages, which I encouraged him to do because I don't have time for
them. That specifically means latin-unity and mule-ucs.
Hrvoje> I also don't know what ISO 8859-13 is, but I'll look into
Hrvoje> it.
It's Latin characters for Baltic Rim languages (Latvian, etc).
However, I'm pretty sure that what's happening here is that it
provides some extra punctuation. Specifically that LOW DOUBLE-9
QUOTATION MARK is unique, apparently, but sophisticated German writers
prefer it to the English raised-style quotation marks.
--
School of Systems and Information Engineering http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
5:14 a.m.
>>>> "Hrvoje" == Hrvoje Niksic
<hniksic(a)xemacs.org> writes:
Hrvoje> "Stephen J. Turnbull" <stephen(a)xemacs.org> writes:
>>>>>> "Jerry" == Jerry James
<james(a)xemacs.org> writes:
Jerry> Stephen, can you comment on Hrvoje's observation?
> Not for some days, maybe weeks.
Hrvoje> Then could we at least fix the bug by applying the
Hrvoje> "obvious" fix I posted? Am I the only one for whom Gnus
Hrvoje> breaks here?
It doesn't break for me. I guess[1] what is desired here is that
(when (eq ?＠ (char-syntax-from-code
(get-range-char-table key new ?＠)))
...)
be changed to
(when (eq ?＠ (char-syntax-from-code
(get-range-char-table key new
(syntax-string-to-code "＠"))))
...)
[[ Can we spell "Ebola", children? Sometimes I'd really like to
disown our parent.... Why couldn't Emacs have been written in Python,
or something designed by Paul Graham? ]] Ahem, back to business.
Does that work for you?
Footnotes:
[1] Literally, I still don't have time to dig in. Use of this might
cause your dog to elope with your PC, so use with caution.
--
School of Systems and Information Engineering http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
7:37 a.m.
>>>> "Jerry" == Jerry James
<james(a)xemacs.org> writes:
Jerry> Listening, yes, but unable to do much more than that right
Jerry> now. While I touched that code last, CVS shows that this
Jerry> is the change that introduced the code Hrvoje is talking
Jerry> about:
Jerry> 2003-07-18 Stephen J. Turnbull <stephen(a)xemacs.org>
Jerry> * derived.el (derived-mode-merge-syntax-tables):
Jerry> Reverse map logic. Fixes "do-nothing" bug.
Something is seriously broken here. That patch was never supposed to
be applied to 21.4 (there is no trouble-making default member in the
representation of 21.4's char-tables), and there's no record of it
being applied to 21.4. Furthermore, CVS thinks that the 21.4 and 21.5
versions are supposed to be identical, and annotates them identically
(as in fact they are in my workspaces). Is that your doing, Jerry, or
is it CVS screwage, or something broken in my workspaces? I can't
convince CVS to show me that change separately in the 21.4 line, how
did you find it? (If you did it inadvertantly, no big deal, but if
CVS is jerking us around ...! <shiver>)
Anyway, reverting that patch (in 21.4 only!) should solve Hrvoje's
problem, and possibly a number of other weirdo bugs related to syntax
tables. It would be a good idea to carefully review the rest of that
file for potential screwage from 21.4/21.5 implementation quirks.
My patch was wrong in the context of 21.5, as well, but it's going to
take more time to figure out how to solve _that_, as a quick look
suggests the necessary Lisp API isn't available.
Vin: I'll post a patch for 21.4 separately.
--
School of Systems and Information Engineering http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
7170
days inactive
7175
days old
15 comments
5 participants
participants (5)
-
Adrian Aichner -
Hrvoje Niksic -
Jerry James -
Stephen J. Turnbull -
Steve Youngs