CVS: Symbol's function definition is void: ignore-errors

CVS: Symbol's function definition is void: ignore-errors

[Thomas Skogestad]

I've never tried CVS before, but I followed the instructions at 
http://www.gnus.org/distribution.html.

(cvs -v gives Concurrent Versions System (CVS) 1.10.6 (client/server))

After downloading (or whatever it's called), I tried starting with
the CVS-Gnus, however just after Gnus has started loading I get the
following error message: Symbol's function definition is void:
ignore-errors  

Just what version is supposed to be downloaded? (gnus.el claims to be
5.8.3)

(Things work just fine with 5.8.2 downloaded from the ftp site.)

If it's of any interest, the debug trace is:

Signaling: (void-function ignore-errors)
  (ignore-errors (nconc (list ... ...) (if ... nil ...)))
  eval((ignore-errors (nconc (list ... ...) (if ... nil ...))))
  custom-initialize-reset(gnus-select-method (ignore-errors (nconc (list ... ...) (if ... nil ...))))
  custom-declare-variable(gnus-select-method (ignore-errors (nconc (list ... ...) (if ... nil ...))) ("/home/thomas/gnus/gnus-5.8.cvs/lisp/gnus.elc" . -20416) :group gnus-server :type gnus-select-method)
  gnus()
  command-line-1(("-f" "gnus"))
  command-line()
  normal-top-level()

-- 
thomas.skogestad@jusstud.uio.no * http://www.ntnu.no/y2k/
http://www.rock103.com/crew/pics/promnight.jpg
http://cgi.mercurycenter.com/premium/comics/10_13/dilbert.gif
http://cgi.mercurycenter.com/premium/comics/11_19/bizarro.gif

Re: CVS: Symbol's function definition is void: ignore-errors

[Raymond Scholz]

Thomas Skogestad <tskogest@jusstud.uio.no> writes:

> After downloading (or whatever it's called), I tried starting with
> the CVS-Gnus, however just after Gnus has started loading I get the
> following error message: Symbol's function definition is void:
> ignore-errors  

I got this error too, it seems to be BBDB related. The backtrace is
the same for me. Uncommenting the following lines in my .gnus solves
the problem temporarily:

;(bbdb-initialize 'gnus 'message 'w3)
;(bbdb-insinuate-message)

The problem came up last night somewhere between 5.8.2 and 5.8.3.

Cheers,
  Ray
-- 
Raymond Scholz -*- rscholz@tzi.de -*- http://www.tzi.de/~rscholz/
FB 3 Informatik, Universität Bremen -*- PGP key available via WWW

Re: CVS: Symbol's function definition is void: ignore-errors

[Andi Hechtbauer]

Raymond Scholz <rscholz@tzi.de> writes:

> Thomas Skogestad <tskogest@jusstud.uio.no> writes:
> 
> > After downloading (or whatever it's called), I tried starting with
> > the CVS-Gnus, however just after Gnus has started loading I get the
> > following error message: Symbol's function definition is void:
> > ignore-errors  
> 
> I got this error too, it seems to be BBDB related. The backtrace is
> the same for me. Uncommenting the following lines in my .gnus solves
> the problem temporarily:
> 
> ;(bbdb-initialize 'gnus 'message 'w3)
> ;(bbdb-insinuate-message)
> 
> The problem came up last night somewhere between 5.8.2 and 5.8.3.

Hmm. I don't use bbdb and still got that error.

About finding out that "ignore-errors" is a macro from "cl-macs",
I manually did a
M-x load-file /usr/share/emacs/20.3/lisp/emacs-lisp/cl-macs.elc
and fired up gnus again: here I am.

There's a (eval-when-compile (require 'cl)) in gnus.el, that perhaps
got something to do with it? OTOH I'm not a lisp hacker at all, I did
do a diff against an ol' pgnus .97 gnus.el, and this line is there,
too...

bye,
-- 
Andi Hechtbauer                                           anti@spin.de 
System Administration                         voice: +49 941 94 65 937
SPiN GmbH               http://www.spin.de/     fax: +49 941 94 65 938  
------- web design - java chats - guestbooks - java/cgi coding -------

Re: CVS: Symbol's function definition is void: ignore-errors

[Michael Welsh Duggan]

Thomas Skogestad <tskogest@jusstud.uio.no> writes:

> I've never tried CVS before, but I followed the instructions at 
> http://www.gnus.org/distribution.html.
> 
> (cvs -v gives Concurrent Versions System (CVS) 1.10.6 (client/server))
> 
> After downloading (or whatever it's called), I tried starting with
> the CVS-Gnus, however just after Gnus has started loading I get the
> following error message: Symbol's function definition is void:
> ignore-errors  

Did you compile the dist?  `ignore-errors' is a macro in the
cl-macs.el package, and calls to it should be expanded at compile
time.

-- 
Michael Duggan
(md5i@cs.cmu.edu)

Re: CVS: Symbol's function definition is void: ignore-errors

[Andi Hechtbauer]

Michael Welsh Duggan <md5i@cs.cmu.edu> writes:

> Thomas Skogestad <tskogest@jusstud.uio.no> writes:
> 
> > I've never tried CVS before, but I followed the instructions at 
> > http://www.gnus.org/distribution.html.
> > 
> > (cvs -v gives Concurrent Versions System (CVS) 1.10.6 (client/server))
> > 
> > After downloading (or whatever it's called), I tried starting with
> > the CVS-Gnus, however just after Gnus has started loading I get the
> > following error message: Symbol's function definition is void:
> > ignore-errors  
> 
> Did you compile the dist?  `ignore-errors' is a macro in the
> cl-macs.el package, and calls to it should be expanded at compile
> time.


make distclean; configure; make

same error.

-- 
Andi Hechtbauer                                           anti@spin.de 
System Administration                         voice: +49 941 94 65 937
SPiN GmbH               http://www.spin.de/     fax: +49 941 94 65 938  
------- web design - java chats - guestbooks - java/cgi coding -------

Re: CVS: Symbol's function definition is void: ignore-errors

[Shenghuo ZHU]

>>>>> "Andi" == Andi Hechtbauer <anti@spin.de> writes:

Andi> Michael Welsh Duggan <md5i@cs.cmu.edu> writes:
>> Thomas Skogestad <tskogest@jusstud.uio.no> writes:
>> 
>> > I've never tried CVS before, but I followed the instructions at 
>> > http://www.gnus.org/distribution.html.
>> > 
>> > (cvs -v gives Concurrent Versions System (CVS) 1.10.6 (client/server))
>> > 
>> > After downloading (or whatever it's called), I tried starting with
>> > the CVS-Gnus, however just after Gnus has started loading I get the
>> > following error message: Symbol's function definition is void:
>> > ignore-errors  
>> 
>> Did you compile the dist?  `ignore-errors' is a macro in the
>> cl-macs.el package, and calls to it should be expanded at compile
>> time.


Andi> make distclean; configure; make

Andi> same error.

[...]

It is a bug. Per Abrahamsen sent a patch to gnus.gnus-bug.

Temporally solution is loading cl-macs.el.

-- 
Shenghuo

Re: CVS: Symbol's function definition is void: ignore-errors

[Per Abrahamsen]

Shenghuo ZHU <zsh@cs.rochester.edu> writes:

> It is a bug. Per Abrahamsen sent a patch to gnus.gnus-bug.

I have committed it now.

Re: CVS: Symbol's function definition is void: ignore-errors

[Emerick Rogul]

Per Abrahamsen writes:

: Shenghuo ZHU <zsh@cs.rochester.edu> writes:
:: It is a bug. Per Abrahamsen sent a patch to gnus.gnus-bug.

: I have committed it now.

Hrm.  I grabbed the most recent gnus out of CVS a few moments ago, and
I'm still having this problem.  Loading cl-macs.el in my .emacs file
seems to bypass the problem, however.  Any ideas what could be
happening?

-Emerick
-- 
-------------------------------------------------------------------------
  "razzle, dazzle, drazzle, drone, time for this one to come home,
   razzle, dazzle, drazzle, dry, time for this one to come alive."
--- Emerick Rogul <emerick@cs.bu.edu> ---------- 'hold my life', the mats

Re: CVS: Symbol's function definition is void: ignore-errors

[Per Abrahamsen]

Emerick Rogul <emerick@cs.bu.edu> writes:

> Per Abrahamsen writes:
> 
> : Shenghuo ZHU <zsh@cs.rochester.edu> writes:
> :: It is a bug. Per Abrahamsen sent a patch to gnus.gnus-bug.
> 
> : I have committed it now.
> 
> Hrm.  I grabbed the most recent gnus out of CVS a few moments ago, and
> I'm still having this problem.  Loading cl-macs.el in my .emacs file
> seems to bypass the problem, however.  Any ideas what could be
> happening?

No.  Try to post a backtrace.

Re: CVS: Symbol's function definition is void: ignore-errors

[Thomas Skogestad]

* Emerick Rogul

| Hrm.  I grabbed the most recent gnus out of CVS a few moments ago,
| and I'm still having this problem.

So did I, just seconds ago. But I still nedd (require 'cl) in my
.emacs.

The backtrace is 

Signaling: (void-function ignore-errors)
  (ignore-errors (nconc (list ... ...) (if ... nil ...)))
  eval((ignore-errors (nconc (list ... ...) (if ... nil ...))))
  custom-initialize-reset(gnus-select-method (ignore-errors (nconc (list ... ...) (if ... nil ...))))
  custom-declare-variable(gnus-select-method (ignore-errors (nconc (list ... ...) (if ... nil ...))) ("/home/thomas/gnus/gnus-5.8.cvs/lisp/gnus.elc" . -20416) :group gnus-server :type gnus-select-method)
  gnus()
  command-line-1(("-f" "gnus"))
  command-line()
  normal-top-level()

-- 
thomas.skogestad@jusstud.uio.no
http://quimby.gnus.org/circus/problem.gif
ftp://ftp.splode.com/pub/users/friedman/emacs-lisp/kill-a-lawyer.el

Re: CVS: Symbol's function definition is void: ignore-errors

[Per Abrahamsen]

Thomas Skogestad <tskogest@jusstud.uio.no> writes:

>   custom-initialize-reset(gnus-select-method (ignore-errors (nconc (list ... ...) (if ... nil ...))))

I missed `gnus-select-method' since I have customized that variable myself.
A fix for that has been committed now.

Re: CVS: Symbol's function definition is void: ignore-errors

[Raymond Scholz]

Per Abrahamsen <abraham@dina.kvl.dk> writes:

> I missed `gnus-select-method' since I have customized that variable myself.
> A fix for that has been committed now.

Hmm, didn't fix it for me. Before your fix I got the backtrace Thomas
reported.

Now it has changed to:

Signaling: (error "Attempted use of a nil select method")
  signal(error ("Attempted use of a nil select method"))
  error("Attempted use of a nil select method")
  gnus-server-opened(nil)
  gnus-start-news-server(nil)
  gnus-1(nil nil nil)
  gnus()
  command-line-1(("-f" "gnus"))
  command-line()
  normal-top-level()

(require 'cl) in .emacs is still required for a successful startup.

Cheers,
  Ray
-- 
Raymond Scholz -*- rscholz@tzi.de -*- http://www.tzi.de/~rscholz/
FB 3 Informatik, Universität Bremen -*- PGP key available via WWW

Re: CVS: Symbol's function definition is void: ignore-errors

[Per Abrahamsen]

Raymond Scholz <rscholz@tzi.de> writes:

> Per Abrahamsen <abraham@dina.kvl.dk> writes:
> 
> > I missed `gnus-select-method' since I have customized that variable myself.
> > A fix for that has been committed now.
> 
> Hmm, didn't fix it for me.

Oops, the definition had nested calls to `ignore-errors', I only
changed the outer layer.

Both calls should be fixed now in CVS.

Re: CVS: Symbol's function definition is void: ignore-errors

[Emerick Rogul]

Per Abrahamsen writes:

: Raymond Scholz <rscholz@tzi.de> writes:
:: Per Abrahamsen <abraham@dina.kvl.dk> writes:
:: 
:: > I missed `gnus-select-method' since I have customized that variable myself.
:: > A fix for that has been committed now.
:: 
:: Hmm, didn't fix it for me.

: Oops, the definition had nested calls to `ignore-errors', I only
: changed the outer layer.

: Both calls should be fixed now in CVS.

That fixed the problem here.  Thanks Per!

-Emerick
-- 
-------------------------------------------------------------------------
Emerick Rogul             /\/     "how young are you, how old am i?
emerick@cs.bu.edu         /\/      let's count the rings around my eyes."
------------------------------------------------- 'i will dare', the mats

Re: CVS: Symbol's function definition is void: ignore-errors

[Dave Love]

At the weekend I sent a bunch of changes for CL problems at runtime
that haven't been applied yet.  (There may be more changes needed
yet).

See an article headed `Re: more CL at runtime fixes' which it seems
you can find in the bugs group nntp+quimby.gnus.org:gnus.gnus-bug.

Re: CVS: Symbol's function definition is void: ignore-errors

[Michael Welsh Duggan]

Andi Hechtbauer <anti@spin.de> writes:

> Michael Welsh Duggan <md5i@cs.cmu.edu> writes:
> 
> > Thomas Skogestad <tskogest@jusstud.uio.no> writes:
> > 
> > > I've never tried CVS before, but I followed the instructions at 
> > > http://www.gnus.org/distribution.html.
> > > 
> > > (cvs -v gives Concurrent Versions System (CVS) 1.10.6 (client/server))
> > > 
> > > After downloading (or whatever it's called), I tried starting with
> > > the CVS-Gnus, however just after Gnus has started loading I get the
> > > following error message: Symbol's function definition is void:
> > > ignore-errors  
> > 
> > Did you compile the dist?  `ignore-errors' is a macro in the
> > cl-macs.el package, and calls to it should be expanded at compile
> > time.
> 
> 
> make distclean; configure; make
> 
> same error.

See my reply to Neil Crellin in the other thread:
<v1tg0xi4ort.fsf@maru.rem.cs.cmu.edu>

-- 
Michael Duggan
(md5i@cs.cmu.edu)

Re: CVS: Symbol's function definition is void: ignore-errors

[Dave Love]

As a workaround, `(require 'cl)' before loading Gnus.

I've sent Lars a bunch of diffs to fix such problems introduced by
requiring cl only at compile time, per Emacs standards.  `copy-list'
and `delete-duplicates' may also bite.  Sorry I didn't catch this
before.

Re: CVS: Symbol's function definition is void: ignore-errors

[Hrvoje Niksic]

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

> I've sent Lars a bunch of diffs to fix such problems introduced by
> requiring cl only at compile time, per Emacs standards.  `copy-list'
> and `delete-duplicates' may also bite.

It would be really nice if GNU Emacs standards could change to at
least allow the major packages to use the CL extensions.

Of course, the code can be written without use of copy-list,
delete-duplicates, etc., but some of the functions provided by cl.el
are non-trivial, and it's a shame to have to duplicate the work that
went into implementing them.

Re: CVS: Symbol's function definition is void: ignore-errors

[Lars Magne Ingebrigtsen]

Hrvoje Niksic <hniksic@iskon.hr> writes:

> Of course, the code can be written without use of copy-list,
> delete-duplicates, etc., but some of the functions provided by cl.el
> are non-trivial, and it's a shame to have to duplicate the work that
> went into implementing them.

Yup.

How about importing all the functions from cl into Gnus, but renaming
them `gnus-*'?  :-)  Should be possible to do with an
`eval-when-compile' in gnus.el and some creativity with `load' and
`mapatoms'.  

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

Re: CVS: Symbol's function definition is void: ignore-errors

[Lars Magne Ingebrigtsen]

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

> How about importing all the functions from cl into Gnus, but renaming
> them `gnus-*'?  :-)

Er; that was a joke, but now I'm serious.  I could just get a list of
all the defuns in cl-*, map over them and do

(fset 'gnus-* (symbol-function *))

Hm.  But how to write out the result during byte-compilation?  We want 
a gnus-cl.elc file that contains the

(fset 'gnus-* #<compiled-function --->)

Can that be done?  Hm.  (Well, not the "#<compiled-function" stuff,
but the same thing.)

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

Re: CVS: Symbol's function definition is void: ignore-errors

[Andreas Schwab]

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

|> Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
|> 
|> > How about importing all the functions from cl into Gnus, but renaming
|> > them `gnus-*'?  :-)
|> 
|> Er; that was a joke, but now I'm serious.  I could just get a list of
|> all the defuns in cl-*, map over them and do
|> 
|> (fset 'gnus-* (symbol-function *))
|> 
|> Hm.  But how to write out the result during byte-compilation?  We want 
|> a gnus-cl.elc file that contains the
|> 
|> (fset 'gnus-* #<compiled-function --->)
|> 
|> Can that be done?  Hm.  (Well, not the "#<compiled-function" stuff,
|> but the same thing.)

(fset 'gnus-* (eval-when-compile (symbol-function '*)))

-- 
Andreas Schwab                                  "And now for something
SuSE Labs                                        completely different."
schwab@suse.de
SuSE GmbH, Schanzäckerstr. 10, D-90443 Nürnberg

Re: CVS: Symbol's function definition is void: ignore-errors

[Hrvoje Niksic]

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

> Hrvoje Niksic <hniksic@iskon.hr> writes:
> 
> > Of course, the code can be written without use of copy-list,
> > delete-duplicates, etc., but some of the functions provided by cl.el
> > are non-trivial, and it's a shame to have to duplicate the work that
> > went into implementing them.
> 
> Yup.
> 
> How about importing all the functions from cl into Gnus, but
> renaming them `gnus-*'?  :-)

Two remarks:

a) There are a lot of them.  You wouldn't want all that in Gnus,
   renamed as gnus-*.

b) Some of them carry interesting meta-information that is not easy to
   transfer, such as instructions for byte-optimizer.  cl.el is
   smarter than it appears -- some of the seemingly cluttered function
   and macro constructs compile to efficient byte-code.  (See the
   various compiler macro definitions in cl-macs.el)

Maybe it would be a better approach to really really beg Stallman to
allow Gnus to use the nice CL thingies.  Gnus is *huge*.  I don't
think anyone will notice the difference.

Re: CVS: Symbol's function definition is void: ignore-errors

[Lars Magne Ingebrigtsen]

Hrvoje Niksic <hniksic@iskon.hr> writes:

> a) There are a lot of them.  You wouldn't want all that in Gnus,
>    renamed as gnus-*.

cl.el and friends isn't that big.  And I only need the functions, not
the macros.

> b) Some of them carry interesting meta-information that is not easy to
>    transfer, such as instructions for byte-optimizer.  cl.el is
>    smarter than it appears -- some of the seemingly cluttered function
>    and macro constructs compile to efficient byte-code.  (See the
>    various compiler macro definitions in cl-macs.el)

Yup.  That's why I want the byte-code, not the functions themselves.

And it can be done.  gnus-cl.el could be a kinda large
`eval-when-compile' form that contained a macro that would return a
big

(progn
  (defaliast 'gnus-delete-duplicates 

#[(cl-seq &rest cl-keys) "byte code removed" [cl-delete-duplicates cl-seq cl-keys nil] 4 596479]

Er.  But the cl functions call other cl functions.  Er.  Uhm.  Hm.
Can I alter the contents of the byte-compiled function to call the
Gnus functions instead?  My, this is getting rather complicated...

> Maybe it would be a better approach to really really beg Stallman to
> allow Gnus to use the nice CL thingies.  Gnus is *huge*.  I don't
> think anyone will notice the difference.

I think it's mainly a name-pollution thing for RMS.  Having
`delete-duplicates' in the namespace will make people believe that's a 
proper Emacs function and stuff.

But I can ask him again.

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

Re: CVS: Symbol's function definition is void: ignore-errors

[Hrvoje Niksic]

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

> > b) Some of them carry interesting meta-information that is not easy to
> >    transfer, such as instructions for byte-optimizer.  cl.el is
> >    smarter than it appears -- some of the seemingly cluttered function
> >    and macro constructs compile to efficient byte-code.  (See the
> >    various compiler macro definitions in cl-macs.el)
> 
> Yup.  That's why I want the byte-code, not the functions themselves.

But the byte-code doesn't have the whole info.  My point is that, for
instance:

(disassemble
  (lambda () (eql x 'bar)))

...disassembles to code that uses `eq' because it's safe.  OTOH, this:

(disassemble
  (lambda () (eql x y)))

...disassembles to code that uses `equal', because EQUAL has its own
byte-code instruction, and it's faster than calling `eql'.

If you do something like (fset 'gnus-eql (indirect-function 'eql)),
you lose this meta-information.  This is a trivial example, but there
are more.

EQL is a trivial example, but there is more where that came from.
"compiler macros" != "Lisp macros".

> #[(cl-seq &rest cl-keys) "byte code removed" [cl-delete-duplicates cl-seq cl-keys nil] 4 596479]
> 
> Er.  But the cl functions call other cl functions.  Er.  Uhm.  Hm.
> Can I alter the contents of the byte-compiled function to call the
> Gnus functions instead?  My, this is getting rather complicated...

Yup.

> > Maybe it would be a better approach to really really beg Stallman
> > to allow Gnus to use the nice CL thingies.  Gnus is *huge*.  I
> > don't think anyone will notice the difference.
> 
> I think it's mainly a name-pollution thing for RMS.  Having
> `delete-duplicates' in the namespace will make people believe that's
> a proper Emacs function and stuff.

Isn't it, really?  It ships with Emacs, it's GPL-ed, it's assigned to
the FSF, it works nicely, it sings and it cooks for you.  If noone is
supposed to use it, why is it there in the first place?

Re: CVS: Symbol's function definition is void: ignore-errors

[Lars Magne Ingebrigtsen]

Hrvoje Niksic <hniksic@iskon.hr> writes:

> > Yup.  That's why I want the byte-code, not the functions themselves.
> 
> But the byte-code doesn't have the whole info.  My point is that, for
> instance:
> 
> (disassemble
>   (lambda () (eql x 'bar)))
> 
> ...disassembles to code that uses `eq' because it's safe.  OTOH, this:
> 
> (disassemble
>   (lambda () (eql x y)))
> 
> ...disassembles to code that uses `equal', because EQUAL has its own
> byte-code instruction, and it's faster than calling `eql'.

Cool.  I didn't know about `define-compiler-macro'.  

> > I think it's mainly a name-pollution thing for RMS.  Having
> > `delete-duplicates' in the namespace will make people believe that's
> > a proper Emacs function and stuff.
> 
> Isn't it, really?  It ships with Emacs, it's GPL-ed, it's assigned to
> the FSF, it works nicely, it sings and it cooks for you.  If noone is
> supposed to use it, why is it there in the first place?

I think people are supposed to decide for themselves whether to load
cl, or something.

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

Re: CVS: Symbol's function definition is void: ignore-errors

[Hrvoje Niksic]

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

> > Isn't it, really?  It ships with Emacs, it's GPL-ed, it's assigned
> > to the FSF, it works nicely, it sings and it cooks for you.  If
> > noone is supposed to use it, why is it there in the first place?
> 
> I think people are supposed to decide for themselves whether to load
> cl, or something.

No offense to Stallman, but that doesn't really make sense.  cl.el is
supposed to be helpful to programmers, not to users.  If no package is
uses it because people should decide whether they want to use it or
not, then why is it there?

But I'm attacking strawmen here -- I don't know the actual Stallman's
position.  I just wish the impressive amount of work invested in cl.el
could be leveraged elsewhere, like in Gnus.

Re: CVS: Symbol's function definition is void: ignore-errors

[Kai Großjohann]

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

> I think it's mainly a name-pollution thing for RMS.  Having
> `delete-duplicates' in the namespace will make people believe that's a 
> proper Emacs function and stuff.

Maybe he would agree to having a cl-safe package which contains all
the CL functions, but with different names cl-foo rather than foo?

I have communicated with somebody about this before,¹ and that person
has asked RMS, and it appears that RMS is somehow opposed to CL on a
non-rational basis.  After all, Scheme (or Guile) is supposed to
become the standard GNU extension language...  And it is not clear
that anything would really be gained by lots of people using CL-isms
where they should have been using Guile-isms, if Emacs Lisp ever gets
transferred to Guile.

But that's only hearsay and rumor and vague memory and all that.

For the record, I think that it is okay for somebody to not /like/
something for irrational reasons.  Happens to all of us, I think.
Some people try to rationalize (is that the right word?), but if you
admit it, it's okay...

kai

¹ Sorry that I forgot your name :-/

-- 
A preposition is not a good thing to end a sentence with.

Re: CVS: Symbol's function definition is void: ignore-errors

[Dave Love]

>>>>> "HN" == Hrvoje Niksic <hniksic@iskon.hr> writes:

 HN> Maybe it would be a better approach to really really beg Stallman
 HN> to allow Gnus to use the nice CL thingies.  Gnus is *huge*.

Surely, though, some of this is meant to be library code (potentially)
used outside Gnus?  It doesn't seem to be a big issue in practice.  I
didn't fix up so much (though there may be some more to do).

Re: CVS: Symbol's function definition is void: ignore-errors

[Dave Love]

>>>>> "HN" == Hrvoje Niksic <hniksic@iskon.hr> writes:

 HN> It would be really nice if GNU Emacs standards could change to at
 HN> least allow the major packages to use the CL extensions.

It's fine to require cl when compiling.

 HN> Of course, the code can be written without use of copy-list,
 HN> delete-duplicates, etc., but some of the functions provided by cl.el
 HN> are non-trivial, 

`delete-duplicates' isn't entirely trivial...

 HN> and it's a shame to have to duplicate the work that went into
 HN> implementing them.

If the case is made that there's a common need for such particular CL
functions they might be moved into the core.  The usual use of `push'
and `pop' is now covered, for instance.

Re: CVS: Symbol's function definition is void: ignore-errors

[Eric Marsden]

>>>>> "Hrvoje" == Hrvoje Niksic <hniksic@iskon.hr> writes:

  Hrvoje> It would be really nice if GNU Emacs standards could change
  Hrvoje> to at least allow the major packages to use the CL
  Hrvoje> extensions.

hear, hear!  Heck, even minor packages :-)

-- 
Eric Marsden

Re: CVS: Symbol's function definition is void: ignore-errors

[David Maslen]

Eric Marsden <emarsden@mail.dotcom.fr> writes:
> >>>>> "Hrvoje" == Hrvoje Niksic <hniksic@iskon.hr> writes:
> 
>   Hrvoje> It would be really nice if GNU Emacs standards could change
>   Hrvoje> to at least allow the major packages to use the CL
>   Hrvoje> extensions.
> 
> hear, hear!  Heck, even minor packages :-)

Isn't that the sort of talk that got the XEmacs project started?

Re: CVS: Symbol's function definition is void: ignore-errors

[Hrvoje Niksic]

Eric Marsden <emarsden@mail.dotcom.fr> writes:

> >>>>> "Hrvoje" == Hrvoje Niksic <hniksic@iskon.hr> writes:
> 
>   Hrvoje> It would be really nice if GNU Emacs standards could change
>   Hrvoje> to at least allow the major packages to use the CL
>   Hrvoje> extensions.
> 
> hear, hear!  Heck, even minor packages :-)

True, but it can be argued that use of cl.el in small packages causes
unnecessary delay in loading time, memory use, etc.  With an
elephant-sized karakondžula like Gnus, cl.el is only a tear-drop in
the ocean of Elisp.  :-)

David Maslen <david.maslen@iname.com> writes:

> Isn't that the sort of talk that got the XEmacs project started?

It is.  But in this case, it's irrelevant, as Gnus is bundled with GNU
Emacs and respects its standards.  It would be, uh, difficult, to have
one version of Gnus under GNU Emacs, and something else under XEmacs.

Incidentally, XEmacs preloads (dumps) the CL extensions.  (Except for
cl-macs.el, which is auto-loaded.)