Re: emacs-23.2-3 and DBus

Re: emacs-23.2-3 and DBus

[Ken Brown]

On 10/4/2010 11:47 PM, nyc4bos@aol.com wrote:
> Thank you for your work on incorporating Emacs and DBus!
>
> Can you tell me how you are testing it?
>
> I tried to use notifiations.el (from the Emacs 24 trunk).
>
> After loading it, Emacs appears to freeze and I see that
> the emacs and winlogin.exe processes consume almost all of
> the CPU.
>
> After about 30 to 45 minutes, I pressed `C-g' and saw the
> message "Back to top level.".
>
> However, Emacs is still frozen and the only thing I could
> do is to kill emacs from outside of emacs.

Please send Cygwin questions to the Cygwin mailing list, not to 
individual maintainers.

I know virtually nothing about D-Bus.  To test it, I simply followed the 
instructions given by Michael Albinus in the thread 
http://cygwin.com/ml/cygwin/2010-08/threads.html#00330 .  Here's a brief 
summary:

1. Start the messagebus service.

2. Start a dbus session (eval `dbus-launch --auto-syntax`) in a shell.

3. echo $DBUS_SESSION_BUS_ADDRESS .

4. In a different shell, set (and export) $DBUS_SESSION_BUS_ADDRESS to 
this value, and run dbus-monitor --session .

5. Start Emacs in the first shell, and load the dbus library.  You 
should see output from dbus-monitor in the other shell, showing that an 
application has started.  It will also tell you the name of that 
application, like ":1.1".

6. In Emacs, evaluate (dbus-get-unique-name :session) .  You should get 
the same name.

7. Register for signals sent by the D-Bus:

(defun my-dbus-signal-handler (&rest args)
   (message "Signal from bus %s received: %s"
            (dbus-event-bus-name last-input-event) args))

(dbus-register-signal
  :system dbus-service-dbus dbus-path-dbus dbus-interface-dbus
  "NameOwnerChanged" 'my-dbus-signal-handler)

(dbus-register-signal
  :session dbus-service-dbus dbus-path-dbus dbus-interface-dbus
  "NameOwnerChanged" 'my-dbus-signal-handler)

8. Now you could open any other D-Bus application, and you should be
notified in Emacs (in the echo area and the *Messages* buffer).  For 
example, you could try dbus-monitor --session , started in the same 
shell where you started Emacs.

I'm afraid that's all I know.  Unless you think there is a 
Cygwin-specific problem, further questions should probably be directed 
to one of the Emacs lists, where you will find people who are 
knowledgeable about D-Bus integration in Emacs.  BTW, there has been 
quite a bit of development of this integration since the release of 
Emacs 23.2.  Some improvements will appear in Emacs 23.3, and some won't 
appear until 24.1.  So if you're trying to use a library from the Emacs 
24 trunk, it's possible that it's simply not compatible with Emacs 23.

Ken

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Michael Albinus]

Ken Brown <kbrown@cornell.edu> writes:

>> I tried to use notifiations.el (from the Emacs 24 trunk).

Could you, please, eval

  (setq dbus-debug t)

before loading the package? This shall give us more information.

> 4. In a different shell, set (and export) $DBUS_SESSION_BUS_ADDRESS to
> this value, and run dbus-monitor --session .

This would be valuable as well: what does dbus-monitor show?

Best regards, Michael.

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Michael Albinus]

Ken Brown <kbrown@cornell.edu> writes:

> So if you're trying to use a library from the Emacs 24 trunk, it's
> possible that it's simply not compatible with Emacs 23.

Under GNU/Linux, I've shortly tested notifications.el (taken from the
trunk) in the emacs-23 branch; it works w/o problem. This does not seem
to be the problem.

I suspect, under cygwin there is no D-Bus service available for
"org.freedesktop.Notifications". However, Emacs shall not be stalled;
usually there is a timeout of 25 seconds for synchronous D-Bus calls.

> Ken

Best regards, Michael.

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Ken Brown]

On 10/5/2010 2:48 PM, Michael Albinus wrote:
> Ken Brown<kbrown@cornell.edu>  writes:
>
>> So if you're trying to use a library from the Emacs 24 trunk, it's
>> possible that it's simply not compatible with Emacs 23.
>
> Under GNU/Linux, I've shortly tested notifications.el (taken from the
> trunk) in the emacs-23 branch; it works w/o problem. This does not seem
> to be the problem.

It doesn't seem to be a problem under Cygwin either.  I just tried 
loading notifications.el, and Emacs didn't freeze.  So I guess we'll 
have to wait for a more detailed problem report from the OP saying 
exactly what he did.

Ken

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Michael Albinus]

Ken Brown <kbrown@cornell.edu> writes:

> It doesn't seem to be a problem under Cygwin either.  I just tried
> loading notifications.el, and Emacs didn't freeze.  So I guess we'll
> have to wait for a more detailed problem report from the OP saying
> exactly what he did.

Maybe you could also try to send a notification:

(notifications-notify :title "Hello world" :body "from Emacs")

> Ken

Best regards, Michael.

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Ken Brown]

On 10/6/2010 12:18 AM, Michael Albinus wrote:
> Ken Brown  writes:
>
>> It doesn't seem to be a problem under Cygwin either.  I just tried
>> loading notifications.el, and Emacs didn't freeze.  So I guess we'll
>> have to wait for a more detailed problem report from the OP saying
>> exactly what he did.
>
> Maybe you could also try to send a notification:
>
> (notifications-notify :title "Hello world" :body "from Emacs")

This results in the error "The name org.freedesktop.Notifications was 
not provided by any .service files", as you predicted.  So I don't think 
it points to any problem with D-BUS support in Emacs.

Thanks for your help.

Ken


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Michael Albinus]

Ken Brown <kbrown@cornell.edu> writes:

>> Maybe you could also try to send a notification:
>>
>> (notifications-notify :title "Hello world" :body "from Emacs")
>
> This results in the error "The name org.freedesktop.Notifications was
> not provided by any .service files", as you predicted.  So I don't
> think it points to any problem with D-BUS support in Emacs.

I wanted to try myself, and I've downloaded emacs-23.2-3 from the curr
branch. When I do start it via "/usr/bin/emacs -Q -l dbus", it results
in the well known frozen state :-(

Switching to the exp branch, there is no other package offer.

"M-x emacs-version" returns "GNU Emacs 23.2.1 (i686-pc-cygwin, GTK+
Version 2.20.1) of 2010-10-02 on laptop". That looks like a recent
build, but ...

The binary has the following checksum:

$ cksum /usr/bin/emacs
658187514 16254990 /usr/bin/emacs

Using a (self-compiled) Emacs 24, there is no freeze.

> Thanks for your help.
>
> Ken

Best regards, Michael.

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Ken Brown]

On 10/6/2010 10:41 AM, Michael Albinus wrote:
> Ken Brown  writes:
>
>>> Maybe you could also try to send a notification:
>>>
>>> (notifications-notify :title "Hello world" :body "from Emacs")
>>
>> This results in the error "The name org.freedesktop.Notifications was
>> not provided by any .service files", as you predicted.  So I don't
>> think it points to any problem with D-BUS support in Emacs.
>
> I wanted to try myself, and I've downloaded emacs-23.2-3 from the curr
> branch. When I do start it via "/usr/bin/emacs -Q -l dbus", it results
> in the well known frozen state :-(

Works fine for me.

> Switching to the exp branch, there is no other package offer.

That's right.  emacs-23.2-3 is current, and there is no experimental 
version.

> The binary has the following checksum:
>
> $ cksum /usr/bin/emacs
> 658187514 16254990 /usr/bin/emacs

That's not actually the binary; it's a symlink that resolves to either 
/usr/bin/emacs-X11.exe or /usr/bin/emacs-nox.exe, depending on whether 
or not you've installed the emacs-X11 package.  Is it possible that you 
upgraded the emacs package but are still using an old emacs-X11?  If 
not, I don't know what the problem could be.  If you can't figure it 
out, you could try giving the command

   cygcheck -svr > cygcheck.out

and sending the resulting cygcheck.out as an attachment.  Maybe I (or 
someone) will be able to spot something.

Ken


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Michael Albinus]

Ken Brown <kbrown@cornell.edu> writes:

>> The binary has the following checksum:
>>
>> $ cksum /usr/bin/emacs
>> 658187514 16254990 /usr/bin/emacs
>
> That's not actually the binary; it's a symlink that resolves to either
> /usr/bin/emacs-X11.exe or /usr/bin/emacs-nox.exe, depending on whether
> or not you've installed the emacs-X11 package.  Is it possible that
> you upgraded the emacs package but are still using an old emacs-X11?
> If not, I don't know what the problem could be.  If you can't figure
> it out, you could try giving the command
>
>   cygcheck -svr > cygcheck.out
>
> and sending the resulting cygcheck.out as an attachment.  Maybe I (or
> someone) will be able to spot something.

I haven't the cygwin notebook with me, next access for me is tomorrow.
I'll check then.

The only difference I could imagine is that I haven't started the system
bus; only the session bus was running. Could you, please, check with
your environment? Maybe there is still a problem.

> Ken

Best regards, Michael.

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Ken Brown]

On 10/6/2010 1:51 PM, Michael Albinus wrote:
> Ken Brown writes:
>
>>> The binary has the following checksum:
>>>
>>> $ cksum /usr/bin/emacs
>>> 658187514 16254990 /usr/bin/emacs
>>
>> That's not actually the binary; it's a symlink that resolves to either
>> /usr/bin/emacs-X11.exe or /usr/bin/emacs-nox.exe, depending on whether
>> or not you've installed the emacs-X11 package.  Is it possible that
>> you upgraded the emacs package but are still using an old emacs-X11?
>> If not, I don't know what the problem could be.  If you can't figure
>> it out, you could try giving the command
>>
>>    cygcheck -svr>  cygcheck.out
>>
>> and sending the resulting cygcheck.out as an attachment.  Maybe I (or
>> someone) will be able to spot something.
>
> I haven't the cygwin notebook with me, next access for me is tomorrow.
> I'll check then.
>
> The only difference I could imagine is that I haven't started the system
> bus; only the session bus was running. Could you, please, check with
> your environment? Maybe there is still a problem.

Yes, that's it.  Emacs does freeze if I load dbus.el and the system bus 
isn't running.  But the problem doesn't occur in a build of Emacs from 
the trunk.  I vaguely recall that someone reported this problem last 
summer, and you fixed it.  But maybe you only fixed it in the trunk, and 
not in the emacs-23 branch.  Do I remember that correctly?

Ken

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Ken Brown]

On 10/6/2010 2:12 PM, Ken Brown wrote:
> On 10/6/2010 1:51 PM, Michael Albinus wrote:
>> Ken Brown writes:
>>
>>>> The binary has the following checksum:
>>>>
>>>> $ cksum /usr/bin/emacs
>>>> 658187514 16254990 /usr/bin/emacs
>>>
>>> That's not actually the binary; it's a symlink that resolves to either
>>> /usr/bin/emacs-X11.exe or /usr/bin/emacs-nox.exe, depending on whether
>>> or not you've installed the emacs-X11 package.  Is it possible that
>>> you upgraded the emacs package but are still using an old emacs-X11?
>>> If not, I don't know what the problem could be.  If you can't figure
>>> it out, you could try giving the command
>>>
>>>     cygcheck -svr>   cygcheck.out
>>>
>>> and sending the resulting cygcheck.out as an attachment.  Maybe I (or
>>> someone) will be able to spot something.
>>
>> I haven't the cygwin notebook with me, next access for me is tomorrow.
>> I'll check then.
>>
>> The only difference I could imagine is that I haven't started the system
>> bus; only the session bus was running. Could you, please, check with
>> your environment? Maybe there is still a problem.
>
> Yes, that's it.  Emacs does freeze if I load dbus.el and the system bus
> isn't running.  But the problem doesn't occur in a build of Emacs from
> the trunk.  I vaguely recall that someone reported this problem last
> summer, and you fixed it.  But maybe you only fixed it in the trunk, and
> not in the emacs-23 branch.  Do I remember that correctly?

Here's what I was thinking of:

   http://debbugs.gnu.org/cgi/bugreport.cgi?bug=6579

Ken

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Michael Albinus]

Ken Brown <kbrown@cornell.edu> writes:

>>> I haven't the cygwin notebook with me, next access for me is tomorrow.
>>> I'll check then.

I couldn't do it today, unfortunately.

>> Yes, that's it.  Emacs does freeze if I load dbus.el and the system bus
>> isn't running.  But the problem doesn't occur in a build of Emacs from
>> the trunk.  I vaguely recall that someone reported this problem last
>> summer, and you fixed it.  But maybe you only fixed it in the trunk, and
>> not in the emacs-23 branch.  Do I remember that correctly?

In the trunk, we have rewritten D-Bus integration into the main
loop. xd_pending_messages, which I believe is the source of our pain,
does not exist any longer in the trunk.

> Here's what I was thinking of:
>
>   http://debbugs.gnu.org/cgi/bugreport.cgi?bug=6579

Yep. Reading the code, I have the feeling, that the following patch
should help:

--8<---------------cut here---------------start------------->8---
*** /home/albinus/src/emacs-23/src/dbusbind.c.~100084~	2010-10-07 16:44:04.632300629 +0200
--- /home/albinus/src/emacs-23/src/dbusbind.c	2010-10-07 16:41:25.000000000 +0200
***************
*** 1571,1577 ****

  /* Check, whether there is pending input in the message queue of the
     D-Bus BUS.  BUS is a Lisp symbol, either :system or :session.  */
! int
  xd_get_dispatch_status (bus)
       Lisp_Object bus;
  {
--- 1571,1577 ----

  /* Check, whether there is pending input in the message queue of the
     D-Bus BUS.  BUS is a Lisp symbol, either :system or :session.  */
! static Lisp_Object
  xd_get_dispatch_status (bus)
       Lisp_Object bus;
  {
***************
*** 1587,1609 ****
    return
      (dbus_connection_get_dispatch_status (connection)
       == DBUS_DISPATCH_DATA_REMAINS)
!     ? TRUE : FALSE;
  }

  /* Check for queued incoming messages from the system and session buses.  */
  int
  xd_pending_messages ()
  {

    /* Vdbus_registered_objects_table will be initialized as hash table
       in dbus.el.  When this package isn't loaded yet, it doesn't make
       sense to handle D-Bus messages.  */
!   return (HASH_TABLE_P (Vdbus_registered_objects_table)
! 	  ? (xd_get_dispatch_status (QCdbus_system_bus)
! 	     || ((getenv ("DBUS_SESSION_BUS_ADDRESS") != NULL)
! 		 ? xd_get_dispatch_status (QCdbus_session_bus)
! 		 : FALSE))
! 	  : FALSE);
  }

  /* Read queued incoming message of the D-Bus BUS.  BUS is a Lisp
--- 1587,1617 ----
    return
      (dbus_connection_get_dispatch_status (connection)
       == DBUS_DISPATCH_DATA_REMAINS)
!     ? Qt : Qnil;
  }

  /* Check for queued incoming messages from the system and session buses.  */
  int
  xd_pending_messages ()
  {
+   int ret;
+   xd_in_read_queued_messages = 1;

    /* Vdbus_registered_objects_table will be initialized as hash table
       in dbus.el.  When this package isn't loaded yet, it doesn't make
       sense to handle D-Bus messages.  */
!   ret = (HASH_TABLE_P (Vdbus_registered_objects_table)
! 	 ? (!NILP (internal_catch (Qdbus_error, xd_get_dispatch_status,
! 				    QCdbus_system_bus))
! 	    || ((getenv ("DBUS_SESSION_BUS_ADDRESS") != NULL)
! 		? (!NILP (internal_catch (Qdbus_error, xd_get_dispatch_status,
! 					  QCdbus_session_bus)))
! 		: FALSE))
! 	 : FALSE);
!
!   /* Return.  */
!   xd_in_read_queued_messages = 0;
!   return ret;
  }

  /* Read queued incoming message of the D-Bus BUS.  BUS is a Lisp
--8<---------------cut here---------------end--------------->8---

At least it doesn't hurt under GNU/Linux :-)

> Ken

Best regards, Michael.

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Ken Brown]

On 10/7/2010 11:03 AM, Michael Albinus wrote:
>> Here's what I was thinking of:
>>
>>    http://debbugs.gnu.org/cgi/bugreport.cgi?bug=6579
>
> Yep. Reading the code, I have the feeling, that the following patch
> should help:
[...]

No, I still get the freeze if I load dbus and the system bus isn't running.

Ken

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Michael Albinus]

Ken Brown <kbrown@cornell.edu> writes:

>> Yep. Reading the code, I have the feeling, that the following patch
>> should help:
> [...]
>
> No, I still get the freeze if I load dbus and the system bus isn't running.

That's strange. Today, I could test under cygwin. I've recompiled the
sources from the emacs-23 branch, and I could reproduce the
problem. After applying the patch, Emacs didn't freeze anymore. What's
the difference to your environment?

I have applied the patch, slightly modified, to the emacs-23
branch; it makes sense anyway. Could you, please, check it again?

> Ken

Best regards, Michael.

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Ken Brown]

On 10/8/2010 8:00 AM, Michael Albinus wrote:
> Ken Brown  writes:
>
>>> Yep. Reading the code, I have the feeling, that the following patch
>>> should help:
>> [...]
>>
>> No, I still get the freeze if I load dbus and the system bus isn't running.
>
> That's strange. Today, I could test under cygwin. I've recompiled the
> sources from the emacs-23 branch, and I could reproduce the
> problem. After applying the patch, Emacs didn't freeze anymore. What's
> the difference to your environment?
>
> I have applied the patch, slightly modified, to the emacs-23
> branch; it makes sense anyway. Could you, please, check it again?

You're right.  I was too impatient yesterday.  What I thought was the 
familiar freeze turns out to be just a delay of several seconds after 
pressing a key before I see a response.  But emacs is still not usable 
that way.  So I think the bottom line is that, at least under Cygwin, 
users should make sure the system bus is running before loading dbus.el. 
  As a practical matter, I would think most D-BUS users would just have 
the messagebus service started automatically, so that this wouldn't be 
an issue; that's what happens by default once you run 
/usr/bin/messagebus-config.

Ken

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: emacs-23.2-3 and DBus

[Michael Albinus]

Ken Brown <kbrown@cornell.edu> writes:

> So I think the bottom line is that, at least under Cygwin, users
> should make sure the system bus is running before loading dbus.el.

That's recommended under any system.

Btw, in order to play with a "real" application, I've tried secrets.el
from the trunk, combined with emacs-23. The precondition is
gnome-keyring, which is installable from cygwin's repository in a recent
version, fortunately.

I've played a little bit with the examples given in secrets.el, using
the session keyring ("collection" in its terminology). It was working
fine.

> Ken

Best regards, Michael.

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple