void-function fill-flowed

void-function fill-flowed

[Norbert Koch]

Hi!

Gnus and XEmacs 21.4 from CVS.  I get the following error when
visiting some articles:


Signaling: (void-function fill-flowed)
  fill-flowed()
  mm-inline-text((#<buffer " *mm*"> ("text/plain" (charset . "us-ascii") (format . "flowed")) 7bit nil nil nil nil nil))
  mm-display-inline((#<buffer " *mm*"> ("text/plain" (charset . "us-ascii") (format . "flowed")) 7bit nil nil nil nil nil))
  mm-display-part((#<buffer " *mm*"> ("text/plain" (charset . "us-ascii") (format . "flowed")) 7bit nil nil nil nil nil) t)
  byte-code("..." [not-attachment type ignored display beg handle string-match throw nil mm-inline-override-p 4 "inline" mm-attachment-override-p mm-automatic-display-p mm-inlinable-p mm-inlined-p mm-automatic-external-display-p t split-string mm-handle-media-type "/" "text" gnus-unbuttonized-mime-type-p gnus-insert-mime-button gnus-article-insert-newline -1 (set-buffer gnus-summary-buffer) ((error)) mm-display-part mm-insert-inline mm-get-part gnus-treat-article bufferp 1 text gnus-article-mime-handle-alist id move gnus-article-mime-handles gnus-newsgroup-charset gnus-newsgroup-ignored-charsets mail-parse-ignored-charsets mail-parse-charset] 7)
  gnus-mime-display-single((#<buffer " *mm*"> ("text/plain" (charset . "us-ascii") (format . "flowed")) 7bit nil nil nil nil nil))
  gnus-mime-display-part((#<buffer " *mm*"> ("text/plain" (charset . "us-ascii") (format . "flowed")) 7bit nil nil nil nil nil))
  gnus-display-mime()
  gnus-article-prepare-display()
  gnus-article-prepare(65404 nil)
  gnus-summary-display-article(65404 nil)
  gnus-summary-select-article(nil nil pseudo)
  gnus-summary-scroll-up(1)
  call-interactively(gnus-summary-scroll-up)


Where should fill-flowed be defined?

Thanks,
norbert.

Re: void-function fill-flowed

[Simon Josefsson]

Norbert Koch <viteno@xemacs.org> writes:

> Hi!
>
> Gnus and XEmacs 21.4 from CVS.  I get the following error when
> visiting some articles:
>
> Signaling: (void-function fill-flowed)
...
>
> Where should fill-flowed be defined?

In flow-fill.el, which now has autoload cookies for those functions.
Did you byte compile and load gnus-load?

Re: void-function fill-flowed

[Katsumi Yamaoka]

>>>>> In <iluu160mhcx.fsf@extundo.com>
>>>>>	Simon Josefsson <jas@extundo.com> wrote:

> Norbert Koch <viteno@xemacs.org> writes:

>> Where should fill-flowed be defined?

> In flow-fill.el, which now has autoload cookies for those functions.
> Did you byte compile and load gnus-load?

If loading of gnus-load.el is indispensable, why doesn't gnus.el
load gnus-load.el? ;-)
-- 
Katsumi Yamaoka <yamaoka@jpl.org>

Re: void-function fill-flowed

[Norbert Koch]

Simon Josefsson <jas@extundo.com> writes:

> In flow-fill.el, which now has autoload cookies for those functions.
> Did you byte compile and load gnus-load?

Yes and yes.  fill-flowed is still not known.  But I've only tested an
fboundp, so that might not autoload the library.

norbert.

Re: void-function fill-flowed

[Simon Josefsson]

Katsumi Yamaoka <yamaoka@jpl.org> writes:

>>>>>> In <iluu160mhcx.fsf@extundo.com>
>>>>>>	Simon Josefsson <jas@extundo.com> wrote:
>
>> Norbert Koch <viteno@xemacs.org> writes:
>
>>> Where should fill-flowed be defined?
>
>> In flow-fill.el, which now has autoload cookies for those functions.
>> Did you byte compile and load gnus-load?
>
> If loading of gnus-load.el is indispensable, why doesn't gnus.el
> load gnus-load.el? ;-)

Good question.  I guess the original reason was that gnus-load.el is
automatically generated, and only contain the things that is normally
dumped into emacs when Gnus is part of emacs, so it isn't something
that Gnus would normally need to require explicitly.

But really, is there a reason we don't (require 'gnus-load) in
gnus.el?  It appears that gnus-load is installed by 'make install',
and it does (provide 'gnus-load), and loading it appear to be a safe
and idempotent operation anyway.

All reference to gnus-load would have to be removed when Gnus is
merged into Emacs CVS, though.

Perhaps someone could try adding (require 'gnus-load) to top-level
gnus.el and see if something explodes.

Opinions?

Re: void-function fill-flowed

[Simon Josefsson]

Norbert Koch <viteno@xemacs.org> writes:

> Simon Josefsson <jas@extundo.com> writes:
>
>> In flow-fill.el, which now has autoload cookies for those functions.
>> Did you byte compile and load gnus-load?
>
> Yes and yes.  fill-flowed is still not known.  But I've only tested an
> fboundp, so that might not autoload the library.

Weird.  For safety, I'm putting back the old explicit autoload
statements to mml.el and mm-view.el.  But it would be useful if you
could try the following in your current emacs/gnus: M-x locate-library
RET gnus-load RET, then load that file and make sure it contain an
autoload statement for flow-fill a'la

;;;### (autoloads (fill-flowed fill-flowed-encode) "flow-fill" "flow-fill.el"
;;;;;;  (16279 27619))
;;; Generated autoloads from flow-fill.el

(autoload (quote fill-flowed-encode) "flow-fill" nil nil nil)

(autoload (quote fill-flowed) "flow-fill" nil nil nil)

Then M-x eval-buff RET on the file, and then check if the fboundp
succeed.

Re: void-function fill-flowed

[Norbert Koch]

Simon Josefsson <jas@extundo.com> writes:

> Weird.  For safety, I'm putting back the old explicit autoload
> statements to mml.el and mm-view.el.  

Thanks.

> But it would be useful if you could try the following in your
> current emacs/gnus: M-x locate-library RET gnus-load RET, then load
> that file and make sure it contain an autoload statement for
> flow-fill a'la
>
> ;;;### (autoloads (fill-flowed fill-flowed-encode) "flow-fill" "flow-fill.el"
> ;;;;;;  (16279 27619))
> ;;; Generated autoloads from flow-fill.el
>
> (autoload (quote fill-flowed-encode) "flow-fill" nil nil nil)
>
> (autoload (quote fill-flowed) "flow-fill" nil nil nil)

It looks slightly different:

;;;***
\f
;;;### (autoloads (fill-flowed fill-flowed-encode) "flow-fill" "lisp/flow-fill.el")

(autoload 'fill-flowed-encode "flow-fill" nil nil nil)

(autoload 'fill-flowed "flow-fill" nil nil nil)


But that's an GNU Emacs vs XEmacs thingy, I s'pose.

> Then M-x eval-buff RET on the file, and then check if the fboundp
> succeed.

Nope, (fboundp 'fill-flowed) => nil.

As you've said, weird.

norbert.

Re: void-function fill-flowed

[Simon Josefsson]

Norbert Koch <viteno@xemacs.org> writes:

> (autoload 'fill-flowed "flow-fill" nil nil nil)
...
>> Then M-x eval-buff RET on the file, and then check if the fboundp
>> succeed.
>
> Nope, (fboundp 'fill-flowed) => nil.

Are you sure gnus-load is loaded entirely?  I recall seeing a
top-level check that rejected loading it again if it was already
loaded, maybe it only exist for XEmacs.  Perhaps the XEmacs Gnus
package have its own gnus-load that is loaded before you have a chance
of loading your own replacement gnus-load.

Otherwise, if (autoload 'FOO "foo") followed by (fboundp 'FOO) doesn't
return t, I believe the emacs is broken.

Re: void-function fill-flowed

[Norbert Koch]

Simon Josefsson <jas@extundo.com> writes:

> Are you sure gnus-load is loaded entirely?  I recall seeing a
> top-level check that rejected loading it again if it was already
> loaded, maybe it only exist for XEmacs.  

Hmm, good question.  Performing eval-buffer twice on the open file is
prompted with an "Already loaded" in the echo area.

> Perhaps the XEmacs Gnus package have its own gnus-load that is
> loaded before you have a chance of loading your own replacement
> gnus-load.

Uh, that would not be nice ... I de-install the package version, so I
only have the CVS around.  Although, according to the load-path the
CVS should come first.

> Otherwise, if (autoload 'FOO "foo") followed by (fboundp 'FOO) doesn't
> return t, I believe the emacs is broken.

Yes.

norbert.

Re: void-function fill-flowed

[Norbert Koch]

Norbert Koch <viteno@xemacs.org> writes:

> Uh, that would not be nice ... I de-install the package version, so I
> only have the CVS around.  Although, according to the load-path the
> CVS should come first.

I've played around with the various bits and pieces and it looks like
everything works fine if I get gnus-load loaded very soon in the
process.  Hmm, so I'd say if there's a possibility to require it in
gnus, you, erm ... cough, could revert the reversion :-?

norbert.

Re: void-function fill-flowed

[Simon Josefsson]

Norbert Koch <viteno@xemacs.org> writes:

> Norbert Koch <viteno@xemacs.org> writes:
>
>> Uh, that would not be nice ... I de-install the package version, so I
>> only have the CVS around.  Although, according to the load-path the
>> CVS should come first.
>
> I've played around with the various bits and pieces and it looks like
> everything works fine if I get gnus-load loaded very soon in the
> process.  Hmm, so I'd say if there's a possibility to require it in
> gnus, you, erm ... cough, could revert the reversion :-?

I think the current code is the safest, but in No Gnus we can remove
the autoload statements in mml-sec.el/mm-view.el.

Adding (require 'gnus-load) to gnus.el still seem useful, but
orthogonal to the flow-fill autoload issue.

Re: void-function fill-flowed

[Ted Zlatanov]

On Thu, 23 Oct 2003, jas@extundo.com wrote:

> Adding (require 'gnus-load) to gnus.el still seem useful, but
> orthogonal to the flow-fill autoload issue.

I think gnus-load.el is also necessary for spam-initialize and
gnus-registry-initialize to work as stated in the manual and on this
group, so please let's add gnus-load.el to the gnus.el requires.

If not, I'll fix the manual to say "you have to also load
gnus-load.el."

Thanks
Ted

Re: void-function fill-flowed

[Simon Josefsson]

Ted Zlatanov <tzz@lifelogs.com> writes:

> On Thu, 23 Oct 2003, jas@extundo.com wrote:
>
>> Adding (require 'gnus-load) to gnus.el still seem useful, but
>> orthogonal to the flow-fill autoload issue.
>
> I think gnus-load.el is also necessary for spam-initialize and
> gnus-registry-initialize to work as stated in the manual and on this
> group, so please let's add gnus-load.el to the gnus.el requires.

Generally, it is required for all autoloaded functions that wasn't
part of the Gnus that came with the emacs that the user happen to use.
Perhaps also for custom types.

> If not, I'll fix the manual to say "you have to also load
> gnus-load.el."

IMHO: End-user should never have to worry about it.  People using
alpha releases can probably be expected to follow additional
instructions, but if it can work without additional steps, that is
better.

Re: void-function fill-flowed

[Steve Youngs]

[-- Attachment #1: Type: text/plain, Size: 916 bytes --]

|--==> "TZ" == Ted Zlatanov <tzz@lifelogs.com> writes:

  TZ> On Thu, 23 Oct 2003, jas@extundo.com wrote:
  >>Adding (require 'gnus-load) to gnus.el still seem useful, but
  >>orthogonal to the flow-fill autoload issue.

  TZ> please let's add gnus-load.el to the gnus.el requires.

Please don't.  It will cause havoc for XEmacs users.  In my own
personal set up I go so far as to delete gnus-load.el.  I can safely
do this because I set Gnus up properly with 'auto-autoloads.el' and
'custom-load.el' files.  I've been meaning to patch the build
procedure so that these two files are created _instead of_
gnus-load.el when building with XEmacs.


-- 
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
|              Ashes to ashes, dust to dust.               |
|      The proof of the pudding, is under the crust.       |
|------------------------------<sryoungs@bigpond.net.au>---|

[-- Attachment #2: Type: application/pgp-signature, Size: 256 bytes --]

Re: void-function fill-flowed

[Simon Josefsson]

Steve Youngs <sryoungs@bigpond.net.au> writes:

> |--==> "TZ" == Ted Zlatanov <tzz@lifelogs.com> writes:
>
>   TZ> On Thu, 23 Oct 2003, jas@extundo.com wrote:
>   >>Adding (require 'gnus-load) to gnus.el still seem useful, but
>   >>orthogonal to the flow-fill autoload issue.
>
>   TZ> please let's add gnus-load.el to the gnus.el requires.
>
> Please don't.  It will cause havoc for XEmacs users.

In what way?  Does putting (require 'gnus-load) in .emacs, which is
what GNUS-NEWS currently recommend, also cause problems?

Re: void-function fill-flowed

[Ted Zlatanov]

On Fri, 24 Oct 2003, sryoungs@bigpond.net.au wrote:

> |--==> "TZ" == Ted Zlatanov <tzz@lifelogs.com> writes:
> 
>   TZ> On Thu, 23 Oct 2003, jas@extundo.com wrote:
>   >>Adding (require 'gnus-load) to gnus.el still seem useful, but
>   >>orthogonal to the flow-fill autoload issue.
> 
>   TZ> please let's add gnus-load.el to the gnus.el requires.
> 
> Please don't.  It will cause havoc for XEmacs users.  In my own
> personal set up I go so far as to delete gnus-load.el.  I can safely
> do this because I set Gnus up properly with 'auto-autoloads.el' and
> 'custom-load.el' files.  I've been meaning to patch the build
> procedure so that these two files are created _instead of_
> gnus-load.el when building with XEmacs.

OK, you're not arguing against gnus-load.el, you're reporting that it
does not work with XEmacs.  Can you fix gnus-load.el so it also works
with XEmacs, delegating to your autoloading files in XEmacs but
working normally with GNU Emacs?

Thanks
Ted

Re: void-function fill-flowed

[Steve Youngs]

[-- Attachment #1: Type: text/plain, Size: 1056 bytes --]

|--==> "SJ" == Simon Josefsson <jas@extundo.com> writes:

  SJ> Steve Youngs <sryoungs@bigpond.net.au> writes:
  >>Please don't.  It will cause havoc for XEmacs users.

  SJ> In what way?  Does putting (require 'gnus-load) in .emacs, which
  SJ> is what GNUS-NEWS currently recommend, also cause problems?

For anyone who uses the XEmacs Gnus package, or anyone like me who
rolls their own auto-autoloads.el, they will see an "Already loaded"
error because of:
        
     `(if (featurep 'gnus-autoloads) (error "Already loaded"))'

XEmacs puts this exact form in 'auto-autoloads.el'.  This file is,
surprise surprise, automatically loaded when XEmacs starts.  Surely
GNU/Emacs has a similar mechanism to handle autoloads.  Or are
'autoloads' in GNU/Emacs really 'manual-loads'?


-- 
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
|              Ashes to ashes, dust to dust.               |
|      The proof of the pudding, is under the crust.       |
|------------------------------<sryoungs@bigpond.net.au>---|

[-- Attachment #2: Type: application/pgp-signature, Size: 256 bytes --]

Re: void-function fill-flowed

[Steve Youngs]

[-- Attachment #1: Type: text/plain, Size: 900 bytes --]

|--==> "TZ" == Ted Zlatanov <tzz@lifelogs.com> writes:

  TZ> On Fri, 24 Oct 2003, sryoungs@bigpond.net.au wrote:
  >>Please don't.  It will cause havoc for XEmacs users.  In my own

[...]

  TZ> OK, you're not arguing against gnus-load.el, you're reporting
  TZ> that it does not work with XEmacs.  Can you fix gnus-load.el so
  TZ> it also works with XEmacs, delegating to your autoloading files
  TZ> in XEmacs but working normally with GNU Emacs?

Sure.  I'll put something together later today.

Um, should this not wait till No Gnus?  Even just putting `(require
'gnus-load)' in gnus.el would be fairly opaque to the user.

-- 
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
|              Ashes to ashes, dust to dust.               |
|      The proof of the pudding, is under the crust.       |
|------------------------------<sryoungs@bigpond.net.au>---|

[-- Attachment #2: Type: application/pgp-signature, Size: 256 bytes --]

Re: void-function fill-flowed

[Ted Zlatanov]

On Fri, 24 Oct 2003, sryoungs@bigpond.net.au wrote:

> |--==> "TZ" == Ted Zlatanov <tzz@lifelogs.com> writes:

>   TZ> OK, you're not arguing against gnus-load.el, you're reporting
>   TZ> that it does not work with XEmacs.  Can you fix gnus-load.el
>   TZ> so it also works with XEmacs, delegating to your autoloading
>   TZ> files in XEmacs but working normally with GNU Emacs?
> 
> Sure.  I'll put something together later today.
> 
> Um, should this not wait till No Gnus?  Even just putting `(require
> 'gnus-load)' in gnus.el would be fairly opaque to the user.

I would argue that because (spam-initialize) by itself does not work
currently as it's supposed to, this will fix a bug and is not a new
feature.  

If anyone else thinks otherwise, please speak up.

Ted

Re: void-function fill-flowed

[Steve Youngs]

[-- Attachment #1: Type: text/plain, Size: 1111 bytes --]

|--==> "TZ" == Ted Zlatanov <tzz@lifelogs.com> writes:

  TZ> On Fri, 24 Oct 2003, sryoungs@bigpond.net.au wrote:
  >>|--==> "TZ" == Ted Zlatanov <tzz@lifelogs.com> writes:

  TZ> OK, you're not arguing against gnus-load.el, you're reporting
  TZ> that it does not work with XEmacs.  Can you fix gnus-load.el
  TZ> so it also works with XEmacs, delegating to your autoloading
  TZ> files in XEmacs but working normally with GNU Emacs?
  >>
  >>Sure.  I'll put something together later today.

In CVS now.

  >>Um, should this not wait till No Gnus?  Even just putting `(require
  >>'gnus-load)' in gnus.el would be fairly opaque to the user.

  TZ> I would argue that because (spam-initialize) by itself does not work
  TZ> currently as it's supposed to, this will fix a bug and is not a new
  TZ> feature.  

OK, I'm cool with that.

-- 
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
|              Ashes to ashes, dust to dust.               |
|      The proof of the pudding, is under the crust.       |
|------------------------------<sryoungs@bigpond.net.au>---|

[-- Attachment #2: Type: application/pgp-signature, Size: 256 bytes --]