No doc strings from DOC

No doc strings from DOC

[Juanma Barranquero]

After the last few changes to doc.c I get, for example:

M-x describe-function set-fill-function

Debugger entered--Lisp error: (wrong-type-argument stringp 0)
  ad-Orig-documentation(set-fill-column nil)
  documentation(set-fill-column)
  describe-function-1(set-fill-column)
  describe-function(set-fill-column)
  call-interactively(describe-function)
  execute-extended-command(nil)
* call-interactively(execute-extended-command)

but I believe that's just uncovering a previous bug (or build problem),
because previously, on my system (Windows XP, compiling with MSVC 6.0),
C-h f did not show documentation for functions in DOC. Perhaps it is
related to the problem in thread "Lost argument and doc string":

http://mail.gnu.org/pipermail/emacs-devel/2002-February/005583.html


                                                           /L/e/k/t/u

Re: No doc strings from DOC

[Juanma Barranquero]

On Wed, 03 Apr 2002 13:10:53 +0200, Juanma Barranquero <lektu@terra.es> wrote:

> C-h f did not show documentation for functions in DOC.

Err... Just the reverse. Functions whose documentation is in DOC are
shown just fine. But, for example:

C-h f complete-symbol           ->  "Wrong type argument, stringp 0"

M-x load-library bindings.elc
C-h f complete-symbol           -> OK

Any idea about how to debug that?

                                                           /L/e/k/t/u

Re: No doc strings from DOC

[Stefan Monnier]

> 
> On Wed, 03 Apr 2002 13:10:53 +0200, Juanma Barranquero <lektu@terra.es> wrote:
> 
> > C-h f did not show documentation for functions in DOC.
> 
> Err... Just the reverse. Functions whose documentation is in DOC are
> shown just fine. But, for example:
> 
> C-h f complete-symbol           ->  "Wrong type argument, stringp 0"

How about a *Backtrace* ?


	Stefan

Re: No doc strings from DOC

[Stefan Monnier]

> After the last few changes to doc.c I get, for example:
> 
> M-x describe-function set-fill-function
> 
> Debugger entered--Lisp error: (wrong-type-argument stringp 0)

Thank you.  I'll make sure the function returns nil instead of 0
in that case.


	Stefan

Re: No doc strings from DOC

[Juanma Barranquero]

On Wed, 03 Apr 2002 11:18:12 -0500, "Stefan Monnier" <monnier+gnu/emacs@RUM.cs.yale.edu> wrote:

> Thank you.  I'll make sure the function returns nil instead of 0
> in that case.

That still produces "Wrong type argument: stringp, 0" if you answer no
to the question about reloading.

But anyway, why do I get "File DOC-X is out-of-sync" even before a full

nmake realclean
configure --with-msvc
nmake bootstrap

?


                                                           /L/e/k/t/u

Re: No doc strings from DOC

[Stefan Monnier]

> 
> On Wed, 03 Apr 2002 11:18:12 -0500, "Stefan Monnier" <monnier+gnu/emacs@RUM.cs.yale.edu> wrote:
> 
> > Thank you.  I'll make sure the function returns nil instead of 0
> > in that case.
> 
> That still produces "Wrong type argument: stringp, 0" if you answer no
> to the question about reloading.

Still with the code I just committed ?

> But anyway, why do I get "File DOC-X is out-of-sync" even before a full

I believe that was fixed yesterday already.
Do you still se that with the current code ?
If so, can you give me a recipe ?


	Stefan

Re: No doc strings from DOC

[Juanma Barranquero]

On Wed, 03 Apr 2002 12:04:57 -0500, "Stefan Monnier" <monnier+gnu/emacs@RUM.cs.yale.edu> wrote:

> Still with the code I just committed ?

Are you talking of

!   if (EQ (tem, make_number (0))
!       tem = Qnil;
!   if (INTEGERP (doc) || CONSP (doc))

and

!   if (EQ (tem, make_number (0))
!       tem = Qnil;
!   if (INTEGERP (tem) || (CONSP (tem) && INTEGERP (XCDR (tem))))

I suppose? Then, the answer is yes.

Without that patch I got an error when doing C-h f set-fill-column, for
example. With it I just get the error if I do C-h f set-fill-column and
answer "no" to the reloading question. If I answer "yes", the help for
the function is displayed, but it says "not documented".

> I believe that was fixed yesterday already.
> Do you still se that with the current code ?

I did my latest cvs update a few minutes ago, and yes, I'm seeing it.

> If so, can you give me a recipe ?

No, I don't know what's happening. Do you think it could be perhaps
related with a CR vs CR/LF issue on .el files? I'm on Windows and
perhaps my CVS client is doing something strange.


                                                           /L/e/k/t/u

Re: No doc strings from DOC

[Stefan Monnier]

> 
> On Wed, 03 Apr 2002 12:04:57 -0500, "Stefan Monnier" <monnier+gnu/emacs@RUM.cs.yale.edu> wrote:
> 
> > Still with the code I just committed ?
> 
> Are you talking of
> 
> !   if (EQ (tem, make_number (0))
> !       tem = Qnil;
> !   if (INTEGERP (doc) || CONSP (doc))
> 
> and
> 
> !   if (EQ (tem, make_number (0))
> !       tem = Qnil;
> !   if (INTEGERP (tem) || (CONSP (tem) && INTEGERP (XCDR (tem))))
> 
> I suppose? Then, the answer is yes.

Huh!

> Without that patch I got an error when doing C-h f set-fill-column, for
> example. With it I just get the error if I do C-h f set-fill-column and
> answer "no" to the reloading question. If I answer "yes", the help for

The question is "why do you get a reloading question" ?

> the function is displayed, but it says "not documented".

But you don't get "Wrong type argument: stringp, 0" any more, right ?
And you did say earlier that you already got "not documented" before
my changes to doc.c, right ?

So the only change left is that you get prompted for reload.
Do I understand it right?

> > If so, can you give me a recipe ?
> No, I don't know what's happening.

You might be right that the problem has to do with EOL.
If you get a reloading question that means that Emacs has a proper
file-offset to the docstring but that when it looks it up it
decides that the offset is bogus.

Could you try to put a breakpoint on get_doc_string and see what's
happening there ?
You could for example see the value of `filepos' and then
look up the DOC-X file to see where that points to and maybe
you'll get some idea about what's wrong.


	Stefan

Re: No doc strings from DOC

[Juanma Barranquero]

On Wed, 03 Apr 2002 12:32:35 -0500, "Stefan Monnier" <monnier+gnu/emacs@RUM.cs.yale.edu> wrote:

> But you don't get "Wrong type argument: stringp, 0" any more, right ?
> And you did say earlier that you already got "not documented" before
> my changes to doc.c, right ?
> 
> So the only change left is that you get prompted for reload.
> Do I understand it right?

After updating from CVS and bootstraping today I no longer get the
reloading question, nor the "Wrong type argument" error. Just a "not
documented" comment in the help.

It is happening for functions in standard modules that get preloaded,
i.e. simple.el, subr.el, widget.el, etc. If I manually reload one of
those files, either in .el or .elc form, the comments are OK.

For the failing functions, the code

  tem = AREF (fun, COMPILED_DOC_STRING);

in Fdocumentation (doc.c, line 396) is returning 0, but I don't know why.
Any more hints on how to debug this?


                                                           /L/e/k/t/u

Re: No doc strings from DOC

[Juanma Barranquero]

On Thu, 04 Apr 2002 10:27:08 +0200, Juanma Barranquero <lektu@terra.es> wrote:

> Any more hints on how to debug this?

Hmmmm...

After:

nmake realclean
nmake bootstrap
nmake install

some functions from preloaded files do not have docstrings (their
bytecode function object has 0 on the DOCSTRING element).

However, I afterwards I do:

nmake clean
nmake
nmake install

and the newly compiled Emacs has the correct DOCSTRING references.

Comparing DOC-X from the bootstrap compilation to the one for the normal
compilation, I see the first one is missing two good-sized chunks.
Looking at the output of the bootstrap compilation I see:

[...]
	"C:\BIN\emacs\devel\lib-src/obj/i386/make-docfile" -d ../src abbrev.c alloc.c alloca.c buffer.c bytecode.c callint.c callproc.c casefiddle.c cm.c cmds.c charset.c coding.c category.c ccl.c data.c dired.c dispnew.c doc.c  doprnt.c editfns.c emacs.c eval.c fileio.c filelock.c filemode.c fns.c fontset.c indent.c insdel.c keyboard.c keymap.c lastfile.c lread.c macros.c marker.c minibuf.c xfaces.c print.c process.c regex.c scroll.c search.c syntax.c sysdep.c term.c termcap.c tparam.c undo.c unexw32.c window.c xdisp.c casetab.c floatfns.c frame.c gmalloc.c intervals.c ralloc.c textprop.c vm-limit.c region-cache.c strftime.c w32.c w32console.c xfaces.c w32fns.c w32heap.c w32inevt.c w32proc.c w32reg.c w32menu.c w32select.c w32term.c w32xfns.c > DOC
	"C:\BIN\emacs\devel\lib-src/obj/i386/make-docfile" -d ../src ../lisp/abbrev.elc  ../lisp/buff-menu.elc  ../lisp/button.elc  ../lisp/byte-run.elc  ../lisp/cus-start.elc  ../lisp/custom.elc  ../lisp/emacs-lisp/backquote.elc  ../lisp/emacs-lisp/lisp-mode.elc  ../lisp/emacs-lisp/lisp.elc  ../lisp/env.elc  ../lisp/faces.elc  ../lisp/files.elc  ../lisp/format.elc  ../lisp/facemenu.elc  ../lisp/select.elc ../lisp/scroll-bar.elc ../lisp/mouse.elc  ../lisp/float-sup.elc  ../lisp/frame.elc ../lisp/help.elc  ../lisp/indent.elc  ../lisp/isearch.elc  ../lisp/loadup.el  ../lisp/loaddefs.el  ../lisp/bindings.elc  ../lisp/map-ynp.elc  ../lisp/menu-bar.elc  ../lisp/international/mule.elc  ../lisp/international/mule-conf.el  ../lisp/international/mule-cmds.elc  ../lisp/international/characters.elc  ../lisp/international/utf-8.elc  ../lisp/international/latin-1.el  ../lisp/international/latin-2.el  ../lisp/international/latin-3.el  ../lisp/international/latin-4.el  ../lisp/international/latin-5.el  ../lisp/international/latin-8.el  ../lisp/international/latin-9.el  ../lisp/case-table.elc  ../lisp/language/chinese.elc  ../lisp/language/cyrillic.elc  ../lisp/language/indian.elc  ../lisp/language/devanagari.elc  ../lisp/language/english.el  ../lisp/language/ethiopic.elc  ../lisp/language/european.elc  ../lisp/language/czech.el  ../lisp/language/slovak.el  ../lisp/language/romanian.el  ../lisp/language/greek.el  ../lisp/language/hebrew.el  ../lisp/language/japanese.el  ../lisp/language/korean.el  ../lisp/language/lao.el  ../lisp/language/thai.el  ../lisp/language/tibetan.elc  ../lisp/language/vietnamese.elc  ../lisp/language/misc-lang.el  ../lisp/language/utf-8-lang.el  ../lisp/language/georgian.el  ../lisp/paths.el  ../lisp/register.elc  ../lisp/replace.elc  ../lisp/simple.elc  ../lisp/startup.elc  ../lisp/subr.elc  ../lisp/term/tty-colors.elc  ../lisp/textmodes/fill.elc  ../lisp/textmodes/page.elc  ../lisp/textmodes/paragraphs.elc  ../lisp/textmodes/text-mode.elc  ../lisp/vc-hooks.elc  ../lisp/ediff-hook.elc  ../lisp/vmsproc.elc ../lisp/vms-patch.elc  ../lisp/dos-fns.elc ../lisp/dos-vars.elc ../lisp/international/ccl.elc ../lisp/international/codepage.elc  ../lisp/ls-lisp.elc ../lisp/disp-table.elc ../lisp/w32-fns.elc ../lisp/dos-w32.elc ../lisp/w32-vars.elc  ../lisp/widget.elc  ../lisp/window.elc  ../lisp/version.el >> DOC
../lisp/abbrev.elc: No such file or directory
../lisp/buff-menu.elc: No such file or directory
../lisp/button.elc: No such file or directory
../lisp/byte-run.elc: No such file or directory
../lisp/cus-start.elc: No such file or directory
../lisp/custom.elc: No such file or directory
../lisp/emacs-lisp/backquote.elc: No such file or directory
../lisp/emacs-lisp/lisp-mode.elc: No such file or directory
../lisp/emacs-lisp/lisp.elc: No such file or directory
../lisp/env.elc: No such file or directory
../lisp/faces.elc: No such file or directory
../lisp/files.elc: No such file or directory
../lisp/format.elc: No such file or directory
../lisp/facemenu.elc: No such file or directory
../lisp/select.elc: No such file or directory
../lisp/scroll-bar.elc: No such file or directory
../lisp/mouse.elc: No such file or directory
../lisp/float-sup.elc: No such file or directory
../lisp/frame.elc: No such file or directory
../lisp/help.elc: No such file or directory
../lisp/indent.elc: No such file or directory
../lisp/isearch.elc: No such file or directory
../lisp/bindings.elc: No such file or directory
../lisp/map-ynp.elc: No such file or directory
../lisp/menu-bar.elc: No such file or directory
../lisp/international/mule.elc: No such file or directory
../lisp/international/mule-cmds.elc: No such file or directory
../lisp/international/characters.elc: No such file or directory
../lisp/international/utf-8.elc: No such file or directory
../lisp/case-table.elc: No such file or directory
../lisp/language/chinese.elc: No such file or directory
../lisp/language/cyrillic.elc: No such file or directory
../lisp/language/indian.elc: No such file or directory
../lisp/language/devanagari.elc: No such file or directory
../lisp/language/ethiopic.elc: No such file or directory
../lisp/language/european.elc: No such file or directory
../lisp/language/tibetan.elc: No such file or directory
../lisp/language/vietnamese.elc: No such file or directory
../lisp/register.elc: No such file or directory
../lisp/replace.elc: No such file or directory
../lisp/simple.elc: No such file or directory
../lisp/startup.elc: No such file or directory
../lisp/subr.elc: No such file or directory
../lisp/term/tty-colors.elc: No such file or directory
../lisp/textmodes/fill.elc: No such file or directory
../lisp/textmodes/page.elc: No such file or directory
../lisp/textmodes/paragraphs.elc: No such file or directory
../lisp/textmodes/text-mode.elc: No such file or directory
../lisp/vc-hooks.elc: No such file or directory
../lisp/ediff-hook.elc: No such file or directory
../lisp/vmsproc.elc: No such file or directory
../lisp/vms-patch.elc: No such file or directory
../lisp/dos-fns.elc: No such file or directory
../lisp/dos-vars.elc: No such file or directory
../lisp/international/ccl.elc: No such file or directory
../lisp/international/codepage.elc: No such file or directory
../lisp/ls-lisp.elc: No such file or directory
../lisp/disp-table.elc: No such file or directory
../lisp/w32-fns.elc: No such file or directory
../lisp/dos-w32.elc: No such file or directory
../lisp/w32-vars.elc: No such file or directory
../lisp/widget.elc: No such file or directory
../lisp/window.elc: No such file or directory

and those are the files whose docstrings aren't found. So obviously
there's a problem in the makefile wrt bootstrapping. Any
idea what that is and how to fix it?


                                                           /L/e/k/t/u

Re: No doc strings from DOC

[Jason Rumney]

Juanma Barranquero <lektu@terra.es> writes:

> On Thu, 04 Apr 2002 10:27:08 +0200, Juanma Barranquero <lektu@terra.es> wrote:
> 
> > Any more hints on how to debug this?
> 
> Hmmmm...
> 
> After:
> 
> nmake realclean
> nmake bootstrap
> nmake install
> 
> some functions from preloaded files do not have docstrings (their
> bytecode function object has 0 on the DOCSTRING element).

I remember a few weeks ago we reverted a change to build the DOC
file from the .el files on Windows, since building with the .el files
does not work.  I think there may have been another part of that
change that we forgot to revert, ie regenerating the DOC file after
the .elc files have been compiled (perhaps in a different makefile
than the other change).


-- 
Jason Rumney

Re: No doc strings from DOC

[Stefan Monnier]

> Juanma Barranquero <lektu@terra.es> writes:
> 
> > On Thu, 04 Apr 2002 10:27:08 +0200, Juanma Barranquero <lektu@terra.es> wrote:
> > 
> > > Any more hints on how to debug this?
> > 
> > Hmmmm...
> > 
> > After:
> > 
> > nmake realclean
> > nmake bootstrap
> > nmake install
> > 
> > some functions from preloaded files do not have docstrings (their
> > bytecode function object has 0 on the DOCSTRING element).
> 
> I remember a few weeks ago we reverted a change to build the DOC
> file from the .el files on Windows, since building with the .el files
> does not work.  I think there may have been another part of that
> change that we forgot to revert, ie regenerating the DOC file after
> the .elc files have been compiled (perhaps in a different makefile
> than the other change).

The DOC file is not necessary for the bootstrap, so it would
be simpler to not generate it before having generated the elc files.


	Stefan

Re: No doc strings from DOC

[Juanma Barranquero]

On Thu, 04 Apr 2002 13:13:36 -0500, "Stefan Monnier" <monnier+gnu/emacs@RUM.cs.yale.edu> wrote:

> The DOC file is not necessary for the bootstrap, so it would
> be simpler to not generate it before having generated the elc files.

(At least on Windows) you're right, and it's the only clean fix I can
see.

Trying to build DOC from the .el files does not work because
make-docfile, as expected, just dumps docstrings starting by
backslash-newline, and most docstrings in preloaded .el files aren't so.

Delaying the building of DOC to after bytecompiling lisp/*.el works
right.

The following patch implements that behavior for Windows. As I'm no
expert on the Emacs bootstrapping and building process, I'll delay
checking it in to see if someone points out any problem with it that
I've overlooked.


                                                           /L/e/k/t/u




Index: makefile.w32-in
===================================================================
RCS file: /cvs/emacs/nt/makefile.w32-in,v
retrieving revision 1.13
diff -u -r1.13 makefile.w32-in
--- makefile.w32-in	14 Dec 2001 18:14:00 -0000	1.13
+++ makefile.w32-in	10 Apr 2002 11:14:12 -0000
@@ -113,21 +113,24 @@
 	cd ..\src
 	$(MAKE) $(MFLAGS) clean
 	cd ..\lib-src
-	$(MAKE) $(MFLAGS) clean DOC
+	$(MAKE) $(MFLAGS) clean
 	cd ..\src
 	$(MAKE) $(MFLAGS) bootstrap
 	$(MAKE) $(MFLAGS) bootstrap-clean
 	cd ..\lisp
 	$(MAKE) $(MFLAGS) bootstrap
+	cd ..\lib-src
+	$(MAKE) $(MFLAGS) DOC
 	cd ..\nt
 
 bootstrap-gmake:
 	$(MAKE) $(MFLAGS) -C ../lisp bootstrap-clean
 	$(MAKE) $(MFLAGS) -C ../src clean
-	$(MAKE) $(MFLAGS) -C ../lib-src clean DOC
+	$(MAKE) $(MFLAGS) -C ../lib-src clean
 	$(MAKE) $(MFLAGS) -C ../src bootstrap
 	$(MAKE) $(MFLAGS) -C ../src bootstrap-clean
 	$(MAKE) $(MFLAGS) -C ../lisp bootstrap
+	$(MAKE) $(MFLAGS) -C ../lib-src DOC
 
 bootstrap-clean: bootstrap-clean-$(MAKETYPE)

Re: No doc strings from DOC

[Stefan Monnier]

> > The DOC file is not necessary for the bootstrap, so it would
> > be simpler to not generate it before having generated the elc files.
> 
> (At least on Windows) you're right, and it's the only clean fix I can see.

It looks good to me.  That's the way the Unix build works.


	Stefan

Re: No doc strings from DOC

[Jason Rumney]

Juanma Barranquero <lektu@terra.es> writes:

> On Thu, 04 Apr 2002 13:13:36 -0500, "Stefan Monnier" <monnier+gnu/emacs@RUM.cs.yale.edu> wrote:
> 
> The following patch implements that behavior for Windows. As I'm no
> expert on the Emacs bootstrapping and building process, I'll delay
> checking it in to see if someone points out any problem with it that
> I've overlooked.

Looks OK to me.

-- 
Jason Rumney