I'm going insane

I'm going insane

[Lars Magne Ingebrigtsen]

Just "upgraded" to Emacs 20.2.93, and promptly found out that the
.newsrc.eld file written by Gnus under Emacs 20 can't be read by Emacs
19.34.  How absolutely...  (I think I'll stop there.)

How does one do this?  Is there a primer on "Oh, please, dear MULE,
don't destroy all my files"?  Don't write Kvinnof\x8e4ngelset to a
file when I want you to write Kvinnofängelset?

(And I wonder how that will turn out.  Splkt!  How come XEmacs' MULE
never had these problems?)

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@ifi.uio.no * Lars Magne Ingebrigtsen

Re: I'm going insane

[Kai Grossjohann]

>>>>> Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:

  > How does one do this?  Is there a primer on "Oh, please, dear MULE,
  > don't destroy all my files"?  Don't write Kvinnof\x8e4ngelset to a
  > file when I want you to write Kvinnofängelset?

I have (set-language-environment "Latin-1") and no problems.  With
Emacs 20.2.  Don't know about the 20.3 pretest thing.

kai
-- 
You ate somebody? -- Just a leg. -- That's terrible! -- Not with mustard.
(Terry Pratchett: Interesting Times)

Re: I'm going insane

[François Pinard]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1153 bytes --]

Lars Magne Ingebrigtsen <larsi@ifi.uio.no> écrit:

> Just "upgraded" to Emacs 20.2.93

You should at least jump to 20.2.97.  A lot of MULE problems, some being
serious, have been corrected since then.  A few (hmm! :-) remain!

I'm currently setting EMACS_UNIBYTE to 1 in my environment, in hope of
temporary limiting the damage and keeping pretesting manageable, as I have
not much time for doing it all very carefully.  Even with this setting,
I still write a lot of reports, and do not find to provide repeatable
test cases I should.  My intent is to wait until things get more stable,
before retrying without the EMACS_UNIBYTE setting.

> (Splkt!  How come XEmacs' MULE never had these problems?)

Hrvoje often sends me friendly teases, trying to convince to jump into the
XEmacs effort, while I'm not even a XEmacs user (yet :-).  So far, I could
bear the needles he throws in my direction.  But now, *now*, Lars, what
will I do?  Do you fully realize you just gave him a big stock of grenades?

-- 
François Pinard                            mailto:pinard@iro.umontreal.ca
Join the free Translation Project!    http://www.iro.umontreal.ca/~pinard

Re: I'm going insane

[Lars Magne Ingebrigtsen]

Kurt Swanson <ksw@dna.lth.se> writes:

> > Just "upgraded" to Emacs 20.2.93,
> 
> You've recompiled the 20.2 version 93 times?!

:-)  No, it's the pretest.  And it's really 20.2.97.1.

> I have:
> 
>   (require 'latin-1)
>   (standard-display-european t)
> 
> in our site-start.el.  Only other problem is 8-bit chars in .el files,
> which is alleviated with:
> 
>   ;;-*-coding: raw-text;-*-
> 
> on the first line.

That's evil.

> So, you watch Australian soap operas about women's prisons on late
> night Swedish television?

It's my favourite tv show.  :-)

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

Re: I'm going insane

[Lars Magne Ingebrigtsen]

Kai Grossjohann <grossjohann@charly.cs.uni-dortmund.de> writes:

> I have (set-language-environment "Latin-1") and no problems.  With
> Emacs 20.2.  Don't know about the 20.3 pretest thing.

It doesn't seem to help - `s' in the group buffer insists on
MULE-ifying .newsrc.eld...

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

Re: I'm going insane

[Hallvard B Furuseth]

Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:

> How does one do this?  Is there a primer on "Oh, please, dear MULE,
> don't destroy all my files"?

Sure.  First you send a bug report.  That's what pretest is for.

Next, never let emacs20 touch a file unless you (or your ISP) have made
a backup of it.  One simple implementation is to put this in ~/.emacs:

(if (fboundp 'coding-system-get)
    (kill-emacs 1))

> How come XEmacs' MULE never had these problems?

Maybe XEmacs doesn't insist that multibyte mode is Just Fine, and allows
the user to turn it off once and for all?  Like this new clever idea to
run bytecomp in multibye mode by default even under -unibyte.

-- 
Hallvard

Re: I'm going insane

[Lars Magne Ingebrigtsen]

Hallvard B Furuseth <h.b.furuseth@usit.uio.no> writes:

> Sure.  First you send a bug report.  That's what pretest is for.

Yup.  I kinda freaked out.

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

Re: I'm going insane

[Michael Harnois]

Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:

> (And I wonder how that will turn out.  Splkt!  How come XEmacs' MULE
> never had these problems?)

The XEmacs god is wreaking her revenge upon you. Repent now, before it 
is too late.

-- 
Michael D. Harnois, Redeemer Lutheran Church, Washburn, IA 
mharnois@sbt.net                      aa0bt@aa0bt.ampr.org 
 A dying people tolerates the present, rejects the future,
 and finds its satisfactions in past greatness and half
 remembered glory. -- John Steinbeck

Re: I'm going insane

[Dave Love]

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

 Lars> Kurt Swanson <ksw@dna.lth.se> writes:
 >> > Just "upgraded" to Emacs 20.2.93,
 >> 
 >> You've recompiled the 20.2 version 93 times?!

 Lars> :-)  No, it's the pretest.  And it's really 20.2.97.1.

 >> I have:
 >> 
 >> (require 'latin-1)
 >> (standard-display-european t)

That's still appropriate in the pretest, though I have the lines in
the opposite order, which probably isn't relevant.  AFAICT both the
EMACS_UNIBYTE environment variable and --unibyte still DTRT.

I no longer have 19.34, but seem able to use a de-MULEtilated 20.2
with .newsrc.eld written by the pretest.  The one thing that I have
seen problems with is gnus-compile; I haven't had the opportunity to
investigate and don't know that it can be blamed on the pretest.
.

Re: I'm going insane

[Hrvoje Niksic]

François Pinard <pinard@iro.umontreal.ca> writes:

> > (Splkt!  How come XEmacs' MULE never had these problems?)
> 
> Hrvoje often sends me friendly teases, trying to convince to jump
> into the XEmacs effort, while I'm not even a XEmacs user (yet :-).
> So far, I could bear the needles he throws in my direction.  But
> now, *now*, Lars, what will I do?  Do you fully realize you just
> gave him a big stock of grenades?

:-)

My teases have mostly resulted from the fact that your mails insisted
on prefixing `ç' with \201 (in octal), which was both incomprehensible 
to anything but MULE-ized Emacs, *and* ugly to behold.

When regarding XEmacs 20 solution of Mule features, there are two
distinct cases one should consider:

1) XEmacs compiled without Mule (which is, in fact, the default) --
   similar to XEmacs 19.16 or GNU Emacs 19.34;

2) XEmacs compiled with Mule.

The fact that XEmacs allows you to do #1 is a Good Thing in and of
itself.  However, the real comparison should be made between GNU Emacs 
20 and case #2.  Some people claim that XEmacs implementation of Mule, 
while somewhat slower, is much cleaner than what GNU Emacs 20.2 has
done.  As far as I can tell from watching the development, GNU Emacs
20.3 is getting to act very similar to XEmacs 20.

-- 
Hrvoje Niksic <hniksic@srce.hr> | Student at FER Zagreb, Croatia
--------------------------------+--------------------------------
Jone's law: The man who smiles when things go wrong has thought of
someone to blame it on.

Re: I'm going insane

[Lars Magne Ingebrigtsen]

Dave Love <d.love@dl.ac.uk> writes:

> I no longer have 19.34, but seem able to use a de-MULEtilated 20.2
> with .newsrc.eld written by the pretest.

Do you have any group names with 8-bit characters?

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

Re: I'm going insane

[François Pinard]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1145 bytes --]

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

> Dave Love <d.love@dl.ac.uk> writes:
> 
> > I no longer have 19.34, but seem able to use a de-MULEtilated 20.2
> > with .newsrc.eld written by the pretest.
> 
> Do you have any group names with 8-bit characters?

I once tried it, while pretesting what became 20.2, and this corrupted
both my nnmail spool, overviews, and `.newsrc.eld', to the point I had
much difficulty and a hard time to recover some good shape out of all this
(and also given I'm far from being familiar with Gnus internals :-).

If I were a good pretester, I know I should try it again, now, and report
problems patiently.  But I confess that I felt miserable enough, the first
time through, that I has been rather shy attempting another try since then.

On the other hand, I would *love* using well-written French names for
mailgroups, and I presume the same applies to a lot of people and languages.
If no one is courageous enough to jump, it will not happen.  Sigh!

-- 
François Pinard                            mailto:pinard@iro.umontreal.ca
Join the free Translation Project!    http://www.iro.umontreal.ca/~pinard

Re: I'm going insane

[Steinar Bang]

>>>>> Lars Magne Ingebrigtsen <larsi@ifi.uio.no>:

> (And I wonder how that will turn out.  Splkt!  How come XEmacs' MULE
> never had these problems?)

Probably because most XEmacsen we see around these parts, are compiled 
without MULE support.

Re: I'm going insane

[Lars Magne Ingebrigtsen]

François Pinard <pinard@iro.umontreal.ca> writes:

> On the other hand, I would *love* using well-written French names for
> mailgroups, and I presume the same applies to a lot of people and languages.
> If no one is courageous enough to jump, it will not happen.  Sigh!

I've now reported it as a bug.  We'll see what happens.

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

Re: I'm going insane

[Lars Magne Ingebrigtsen]

Steinar Bang <sb@metis.no> writes:

> > (And I wonder how that will turn out.  Splkt!  How come XEmacs' MULE
> > never had these problems?)
> 
> Probably because most XEmacsen we see around these parts, are compiled 
> without MULE support.

No, I've run XEmacs 20 with MULE support, and I had none of these
problems.  Well, I can't recall any such problems, at least.

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

Re: I'm going insane

[Dave Love]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 684 bytes --]


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

 Lars> Dave Love <d.love@dl.ac.uk> writes:

 >> I no longer have 19.34, but seem able to use a de-MULEtilated 20.2
 >> with .newsrc.eld written by the pretest.

 Lars> Do you have any group names with 8-bit characters?

I just created `nnvirtual:voilà' with the pretest version in unibyte
mode and read it OK in both the pretest and the MBSK 20.2 afterwards.
Since it appears in .newsrc.eld clearly as unibyte, I wouldn't expect
otherwise.  If I operate in multibyte mode I expect Emacs 19 or Emacs
20/MBSK to lose on the created files, but I don't do that.

What problem should I be trying to reproduce, exactly?
.

Re: I'm going insane

[Hrvoje Niksic]

Dave Love <d.love@dl.ac.uk> writes:

> >>>>> "Lars" == Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
> 
>  Lars> Dave Love <d.love@dl.ac.uk> writes:
> 
>  >> I no longer have 19.34, but seem able to use a de-MULEtilated 20.2
>  >> with .newsrc.eld written by the pretest.
> 
>  Lars> Do you have any group names with 8-bit characters?
> 
> I just created `nnvirtual:voilŕ' with the pretest version in unibyte
> mode and read it OK in both the pretest and the MBSK 20.2 afterwards.
> Since it appears in .newsrc.eld clearly as unibyte, I wouldn't expect
> otherwise.  If I operate in multibyte mode I expect Emacs 19 or Emacs
> 20/MBSK to lose on the created files, but I don't do that.

Why, BTW?  Why not write out latin1 characters as they are?

-- 
Hrvoje Niksic <hniksic@srce.hr> | Student at FER Zagreb, Croatia
--------------------------------+--------------------------------
We must all confront and destroy our own inner geek.

Re: I'm going insane

[Dave Love]

>>>>> "Rat" == Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

 Rat> Don't use MULE.  

Reasonable; I don't use multibyte, at least.

 Rat> Don't use Emacs 20 until it is totally MULE-free.

Foo.  Non sequitur alert.
.

Re: I'm going insane

[Lars Magne Ingebrigtsen]

Dave Love <d.love@dl.ac.uk> writes:

> What problem should I be trying to reproduce, exactly?

If I have a group called "nnml:voilà" (created with Emacs 19.34), then
starting up emacs-20.2.97 with --unibyte, `M-x gnus' and `C-u s' will
write out some MULE atrocity instead of "à" in the .newsrc.eld file.

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

Re: I'm going insane

[Michael Harnois]

Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

> Most probably because MULE in XEmacs was done right, whereas MULE in FSF
> Emacs is the ugliest hack in existence.

Why, that's straightforward! I like straightforward.

-- 
Michael D. Harnois, Redeemer Lutheran Church, Washburn, IA 
mharnois@sbt.net                      aa0bt@aa0bt.ampr.org 
 A dying people tolerates the present, rejects the future,
 and finds its satisfactions in past greatness and half
 remembered glory. -- John Steinbeck