Symbol's function definition is void: x-register-dnd-atom

Symbol's function definition is void: x-register-dnd-atom

[Óscar Fuentes]

On a custom Emacs build (made to avoid dumping) invoking the executable
ends with that message after loading a bunch of files:

<... lots elided ...>
Loading tooltip...
Loading tooltip...done
Finding pointers to doc strings...
Finding pointers to doc strings...done
Symbol's function definition is void: x-register-dnd-atom

x-register-dnd-atom is defined in xselect.c, which is included in the
build. It is configured with USE_X_TOOLKIT USE_LUCID

Any hints about why the message appears?

Any other suggestion about how to make a functional Emacs binary without
dumping would be useful too. (The build already sets CANNOT_DUMP)

Running the executable with -nw fails too:

$ src/qemacs -Q -nw
emacs: Cannot open terminfo database file

On a normal build temacs starts fine in terminal mode.

Re: Symbol's function definition is void: x-register-dnd-atom

[Jan Djärv]

Óscar Fuentes skrev 2010-07-25 05.01:
> On a custom Emacs build (made to avoid dumping) invoking the executable
> ends with that message after loading a bunch of files:
>
> <... lots elided ...>
> Loading tooltip...
> Loading tooltip...done
> Finding pointers to doc strings...
> Finding pointers to doc strings...done
> Symbol's function definition is void: x-register-dnd-atom
>
> x-register-dnd-atom is defined in xselect.c, which is included in the
> build. It is configured with USE_X_TOOLKIT USE_LUCID
>
> Any hints about why the message appears?

It seems as syms_of_xselect has not been run.

>
> Any other suggestion about how to make a functional Emacs binary without
> dumping would be useful too. (The build already sets CANNOT_DUMP)
>
> Running the executable with -nw fails too:
>
> $ src/qemacs -Q -nw
> emacs: Cannot open terminfo database file
>
> On a normal build temacs starts fine in terminal mode.

Did you link with any termcap/terminfo/ncurses library?

	Jan D.

Re: Symbol's function definition is void: x-register-dnd-atom

[Óscar Fuentes]

Jan Djärv <jan.h.d@swipnet.se> writes:

>> Any hints about why the message appears?
>
> It seems as syms_of_xselect has not been run.

Right. After enabling HAVE_X11 it works. Thanks!

BTW, it seems to me that there are quite a few broken
configurations. This is an instance: if HAVE_X11 is not defined, the
problem described on the subject arises.

>> $ src/qemacs -Q -nw
>> emacs: Cannot open terminfo database file
>>
>> On a normal build temacs starts fine in terminal mode.
>
> Did you link with any termcap/terminfo/ncurses library?

Yes, ncurses. termcap is symlinked to ncurses on this system.

Re: Symbol's function definition is void: x-register-dnd-atom

[Jan D.]

Óscar Fuentes skrev 2010-07-25 17:33:
> Jan Djärv<jan.h.d@swipnet.se>  writes:
>
>>> Any hints about why the message appears?
>>
>> It seems as syms_of_xselect has not been run.
>
> Right. After enabling HAVE_X11 it works. Thanks!
>
> BTW, it seems to me that there are quite a few broken
> configurations. This is an instance: if HAVE_X11 is not defined, the
> problem described on the subject arises.

It is kind of assumed that some defines belong together, such as 
USE_LUCID can't be used unless HAVE_X11 is defined, ditto for 
USE_X_TOOLKIT.  I guess you aren't running configure, if you where this 
wouldn't happen (unless configure itself is messed up).

>
>>> $ src/qemacs -Q -nw
>>> emacs: Cannot open terminfo database file
>>>
>>> On a normal build temacs starts fine in terminal mode.
>>
>> Did you link with any termcap/terminfo/ncurses library?
>
> Yes, ncurses. termcap is symlinked to ncurses on this system.
>

You have to debug that I'm afraid.

	Jan D.

Re: Symbol's function definition is void: x-register-dnd-atom

[Óscar Fuentes]

"Jan D." <jan.h.d@swipnet.se> writes:

>> BTW, it seems to me that there are quite a few broken
>> configurations. This is an instance: if HAVE_X11 is not defined, the
>> problem described on the subject arises.
>
> It is kind of assumed that some defines belong together, such as
> USE_LUCID can't be used unless HAVE_X11 is defined, ditto for
> USE_X_TOOLKIT.  I guess you aren't running configure, if you where
> this wouldn't happen (unless configure itself is messed up).

I'm using cmake and the configure tests is work in progress. Still, I
thought that HAVE_X11 could be undefined even when USE_LUCID is
on. Seems I was wrong.

[snip]

Re: Symbol's function definition is void: x-register-dnd-atom

[Eli Zaretskii]

> From: Óscar Fuentes <ofv@wanadoo.es>
> Date: Sun, 25 Jul 2010 05:01:02 +0200
> 
> Any other suggestion about how to make a functional Emacs binary without
> dumping would be useful too.

Try

  $ src/qemacs -Q -loadup

Re: Symbol's function definition is void: x-register-dnd-atom

[Eli Zaretskii]

> From: Óscar Fuentes <ofv@wanadoo.es>
> Date: Sun, 25 Jul 2010 05:01:02 +0200
> 
> Any other suggestion about how to make a functional Emacs binary without
> dumping would be useful too.

Try

  $ src/qemacs -Q -l loadup

(The previous message was in error, sorry.)

Re: Symbol's function definition is void: x-register-dnd-atom

[Óscar Fuentes]

Eli Zaretskii <eliz@gnu.org> writes:

>> Any other suggestion about how to make a functional Emacs binary
>> without dumping would be useful too.
>
> Try
>
>   $ src/qemacs -Q -loadup

Emacs starts and finally complains on the minibuffer:

Unknown option -loadup

Same for --loadup

The most similar option displayed by `emacs --help' is `--load FILE',
but I think it is not the same at all.

Anyways, Emacs starts fine in GUI mode now, so this is not a pressing
issue. I'm interested on possible issues I could find while running a
non-dumped emacs, though.

Re: Symbol's function definition is void: x-register-dnd-atom

[Óscar Fuentes]

Eli Zaretskii <eliz@gnu.org> writes:

>> Any other suggestion about how to make a functional Emacs binary
>> without dumping would be useful too.
>
> Try
>
>   $ src/qemacs -Q -l loadup

Thanks. emacs starts and displays on the minibuffer

byte-code: Category `a' is already defined

It seems to work fine, though.

I will experiment with that option. Something useful would be to not
have the requirement of having the byte-compiled .el files around for
starting Emacs (at some point while starting without .elc files, Emacs
complains about `(require 'cl) while dumping' or something similar, and
stops.)

Re: Symbol's function definition is void: x-register-dnd-atom

[Eli Zaretskii]

> From: Óscar Fuentes <ofv@wanadoo.es>
> Date: Sun, 25 Jul 2010 19:33:39 +0200
> 
> I'm interested on possible issues I could find while running a
> non-dumped emacs, though.

With or without "-l loadup" on the command line?

Without it, you will miss a lot of features which you are accustomed
to have.

With it -- I'm afraid you are more or less on your own, since I doubt
that anyone runs Emacs like that too often.  I only know this trick
because I needed to debug problems that appear due to dumping.

Re: Symbol's function definition is void: x-register-dnd-atom

[Óscar Fuentes]

Eli Zaretskii <eliz@gnu.org> writes:

>> I'm interested on possible issues I could find while running a
>> non-dumped emacs, though.
>
> With or without "-l loadup" on the command line?
>
> Without it, you will miss a lot of features which you are accustomed
> to have.

For the time being, I'm configuring src/epaths.in to point to a
pre-built lisp/ directory. With that, Emacs seems fully functional
without dumping.

> With it -- I'm afraid you are more or less on your own, since I doubt
> that anyone runs Emacs like that too often.  I only know this trick
> because I needed to debug problems that appear due to dumping.

Okay. Thanks.

Re: Symbol's function definition is void: x-register-dnd-atom

[Ken Raeburn]

On Jul 25, 2010, at 14:10, Eli Zaretskii wrote:
>> From: Óscar Fuentes <ofv@wanadoo.es>
>> Date: Sun, 25 Jul 2010 19:33:39 +0200
>> 
>> I'm interested on possible issues I could find while running a
>> non-dumped emacs, though.
> 
> With or without "-l loadup" on the command line?
> 
> Without it, you will miss a lot of features which you are accustomed
> to have.

Óscar's original message said he's using CANNOT_DUMP.  Unless something's been broken since I last tried it, loading loadup.el is part of the program startup in that mode, as long as "--no-loadup"/"-nl" isn't given; it doesn't need to be specified on the command line.  If passing "-l loadup" makes things work better, there's probably a bug in there somewhere...

Ken

Re: Symbol's function definition is void: x-register-dnd-atom

[Óscar Fuentes]

Ken Raeburn <raeburn@raeburn.org> writes:

>> With or without "-l loadup" on the command line?
>> 
>> Without it, you will miss a lot of features which you are accustomed
>> to have.
>
> Óscar's original message said he's using CANNOT_DUMP.  Unless
> something's been broken since I last tried it, loading loadup.el is
> part of the program startup in that mode, as long as
> "--no-loadup"/"-nl" isn't given; it doesn't need to be specified on
> the command line. 

Right. Once compiled with CANNOT_DUMP, if Emacs cannot find loadup.el on
startup it complains and stops.

> If passing "-l loadup" makes things work better, there's probably a
> bug in there somewhere...

I think that using "-l loadup" will save me of hacking around with
src/epaths.el

Re: Symbol's function definition is void: x-register-dnd-atom

[Eli Zaretskii]

> From: Óscar Fuentes <ofv@wanadoo.es>
> Date: Sun, 25 Jul 2010 21:47:01 +0200
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> >> I'm interested on possible issues I could find while running a
> >> non-dumped emacs, though.
> >
> > With or without "-l loadup" on the command line?
> >
> > Without it, you will miss a lot of features which you are accustomed
> > to have.
> 
> For the time being, I'm configuring src/epaths.in to point to a
> pre-built lisp/ directory. With that, Emacs seems fully functional
> without dumping.

If you don't use "-l loadup", some commands and features might fail,
because preloaded Lisp files don't normally have all those autoload
cookies.  So it is safer to load loadup.

Re: Symbol's function definition is void: x-register-dnd-atom

[Eli Zaretskii]

> From: Ken Raeburn <raeburn@raeburn.org>
> Date: Sun, 25 Jul 2010 16:33:59 -0400
> 
> > With or without "-l loadup" on the command line?
> > 
> > Without it, you will miss a lot of features which you are accustomed
> > to have.
> 
> Óscar's original message said he's using CANNOT_DUMP.  Unless something's been broken since I last tried it, loading loadup.el is part of the program startup in that mode, as long as "--no-loadup"/"-nl" isn't given; it doesn't need to be specified on the command line.  If passing "-l loadup" makes things work better, there's probably a bug in there somewhere...

In case it wasn't clear, I was trying to say that there should be no
need to futz with CANNOT_DUMP, especially since almost no platforms
use it nowadays, and consequently few people are familiar with what it
does to help Óscar if something doesn't work.