From erik at nittin.net Sat Mar 1 18:11:02 2003
From: erik at nittin.net (Erik Enge)
Date: Sat Mar 1 18:11:02 2003
Subject: Make `eclipse` executable after `make`.
Message-ID: <877kbjjo3j.fsf@nittin.net>
How about making `eclipse` executable? I'm not sure where you want it
in the Makefile, but perhaps just after core. I'm using Bash as my
shell and after I do `make` I have to do `chmod a+x eclipse` before I
can do `./eclipse :1.0`.
Erik.
From erik at nittin.net Sun Mar 2 05:09:01 2003
From: erik at nittin.net (Erik Enge)
Date: Sun Mar 2 05:09:01 2003
Subject: Try loading ~/.eclipse/custom.lisp before ~/.eclipse
Message-ID: <873cm7jlb5.fsf@nittin.net>
How about this?:
(defun load-configuration ()
"Load the configuration file. If ~/.eclipse/custom.lisp is a file,
load it, if not, try ~/.eclipse and if that fails try the .eclipse
that shipped with the Eclipse code"
(or (load-config-file (home-subdirectory ".eclipse/custom.lisp"))
(load-config-file (home-subdirectory ".eclipse"))
(load-config-file (eclipse-path ".eclipse"))
(format *error-output* "Unable to read a configuration file.~%")
(%quit%)))
And then just call LOAD-CONFIGURATION in INITIALIZE.
I'd like to have my ~/.eclipse load other files and having them all in
one directory would help a lot. I use ~/.xemacs/custom.el too.
I can make this into a patch with `diff` if you want me to? How do you
prefer `diff` to be run? Some people are picky about -u or -c and other
switches.
Erik.
From hatchond at yahoo.fr Sun Mar 2 05:40:02 2003
From: hatchond at yahoo.fr (Iban)
Date: Sun Mar 2 05:40:02 2003
Subject: Try loading ~/.eclipse/custom.lisp before ~/.eclipse
References: <873cm7jlb5.fsf@nittin.net>
Message-ID: <3E621719.4070603@yahoo.fr>
Erik Enge wrote:
> How about this?:
>
> (defun load-configuration ()
> "Load the configuration file. If ~/.eclipse/custom.lisp is a file,
> load it, if not, try ~/.eclipse and if that fails try the .eclipse
> that shipped with the Eclipse code"
> (or (load-config-file (home-subdirectory ".eclipse/custom.lisp"))
> (load-config-file (home-subdirectory ".eclipse"))
> (load-config-file (eclipse-path ".eclipse"))
> (format *error-output* "Unable to read a configuration file.~%")
> (%quit%)))
>
> And then just call LOAD-CONFIGURATION in INITIALIZE.
>
> I'd like to have my ~/.eclipse load other files and having them all in
> one directory would help a lot. I use ~/.xemacs/custom.el too.
>
Well, it looks like a good idea. What does other think about it ?
> I can make this into a patch with `diff` if you want me to? How do you
> prefer `diff` to be run? Some people are picky about -u or -c and other
> switches.
I'm not so familiar with the diff switches, but Christian Lynbech sends
some goods one, so maybe he can tell us his way to do that.
I have still one patch of him to deal with, I'll do it today.
!ban.
___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en fran?ais !
Yahoo! Mail : http://fr.mail.yahoo.com
From erik at nittin.net Sun Mar 2 07:54:01 2003
From: erik at nittin.net (Erik Enge)
Date: Sun Mar 2 07:54:01 2003
Subject: Eclispe website/documentation proposal.
Message-ID: <87r89qi62u.fsf@nittin.net>
I thought I would start by learning how to make some themes and
customize bindings before I really dive into Eclipse hackery. Except
for the man-page, there isn't much documentation, so I thought I'd write
a Themers Manual that describes how themes are organized in Eclipse and
how to make new ones (I'm currently porting AquaOS from sawfish). Then
I thought I'd move on to a Users Manual that describes installation and
customization before finally doing a Programmers Manual which details
how Eclipse's internals are laid out and how to develop it using Emacs,
ILISP and perhaps other tools.
But then I thought, where do we put these manuals once I've written
them? We need a website to put Eclipse stuff on. What does the
community feel about ?
I'm writing Themers Manual right now. I'm using plain LaTeX; does
anybody object to that?
Erik.
From christian at defun.dk Sun Mar 2 08:07:00 2003
From: christian at defun.dk (Christian Lynbech)
Date: Sun Mar 2 08:07:00 2003
Subject: Try loading ~/.eclipse/custom.lisp before ~/.eclipse
In-Reply-To: <3E621719.4070603@yahoo.fr> (Iban's message of "Sun, 02 Mar
2003 15:37:13 +0100")
References: <873cm7jlb5.fsf@nittin.net> <3E621719.4070603@yahoo.fr>
Message-ID: <87u1el1yvu.fsf@baguette.defun.dk>
>>>>> "Iban" == Iban writes:
>> I can make this into a patch with `diff` if you want me to? How do you
>> prefer `diff` to be run? Some people are picky about -u or -c and other
>> switches.
Iban> I'm not so familiar with the diff switches, but Christian Lynbech
Iban> sends some goods one, so maybe he can tell us his way to do that.
I used unidiff (-u), since I find it more readable than context diffs (-c).
I have no idea whether there could be any other reasons to prefer one
over the other.
------------------------+-----------------------------------------------------
Christian Lynbech | email: christian at defun.dk
------------------------+-----------------------------------------------------
Hit the philistines three times over the head with the Elisp reference manual.
- petonic at hal.com (Michael A. Petonic)
From christian at defun.dk Sun Mar 2 08:14:00 2003
From: christian at defun.dk (Christian Lynbech)
Date: Sun Mar 2 08:14:00 2003
Subject: Try loading ~/.eclipse/custom.lisp before ~/.eclipse
In-Reply-To: <873cm7jlb5.fsf@nittin.net> (Erik Enge's message of "Sat, 01
Mar 2003 07:58:06 -0500")
References: <873cm7jlb5.fsf@nittin.net>
Message-ID: <87r89p1ykz.fsf@baguette.defun.dk>
>>>>> "Erik" == Erik Enge writes:
Erik> I'd like to have my ~/.eclipse load other files and having them all in
Erik> one directory would help a lot. I use ~/.xemacs/custom.el too.
It is a neat feature, but there would probably be many more sensible
ways of organising the startup files than we want to code up inside
eclipse.
How about making the default config file name be a configuration
option and perhaps adding a switch to the `eclipse' script that allows
a per-session specification to be made?
I think I could code that up pretty easily (but lets agree on how to
do it first :-)
------------------------+-----------------------------------------------------
Christian Lynbech | email: christian at defun.dk
------------------------+-----------------------------------------------------
Hit the philistines three times over the head with the Elisp reference manual.
- petonic at hal.com (Michael A. Petonic)
From christian at defun.dk Sun Mar 2 08:23:00 2003
From: christian at defun.dk (Christian Lynbech)
Date: Sun Mar 2 08:23:00 2003
Subject: Eclispe website/documentation proposal.
In-Reply-To: <87r89qi62u.fsf@nittin.net> (Erik Enge's message of "Sun, 02
Mar 2003 02:24:41 -0500")
References: <87r89qi62u.fsf@nittin.net>
Message-ID: <87of4t1y59.fsf@baguette.defun.dk>
>>>>> "Erik" == Erik Enge writes:
Erik> But then I thought, where do we put these manuals once I've written
Erik> them? We need a website to put Eclipse stuff on. What does the
Erik> community feel about ?
Sounds great. Even if there will eventually be an official eclipse
site, there is no reason for you to not just publish it in any
convenient way.
Once you have something to show, you should add a record at Cliki so
other may find it.
Erik> I'm writing Themers Manual right now. I'm using plain LaTeX; does
Erik> anybody object to that?
I would personally consider TexInfo. It is very nice for writing
documentation and I find info pages easier to use.
However, if you like LaTeX better or does not know TexInfo, you should
keep going with LaTeX.
------------------------+-----------------------------------------------------
Christian Lynbech | email: christian at defun.dk
------------------------+-----------------------------------------------------
Hit the philistines three times over the head with the Elisp reference manual.
- petonic at hal.com (Michael A. Petonic)
From erik at nittin.net Sun Mar 2 08:45:01 2003
From: erik at nittin.net (Erik Enge)
Date: Sun Mar 2 08:45:01 2003
Subject: Try loading ~/.eclipse/custom.lisp before ~/.eclipse
References: <873cm7jlb5.fsf@nittin.net> <87r89p1ykz.fsf@baguette.defun.dk>
Message-ID: <87of4ui3r0.fsf@nittin.net>
Christian Lynbech writes:
> It is a neat feature, but there would probably be many more sensible
> ways of organising the startup files than we want to code up inside
> eclipse.
Good point.
> How about making the default config file name be a configuration
> option
Agreed.
> and perhaps adding a switch to the `eclipse' script that allows a
> per-session specification to be made?
Sounds good. :)
Erik.
From hatchond at yahoo.fr Sun Mar 2 09:33:00 2003
From: hatchond at yahoo.fr (Iban)
Date: Sun Mar 2 09:33:00 2003
Subject: Eclispe website/documentation proposal.
References: <87r89qi62u.fsf@nittin.net> <87of4t1y59.fsf@baguette.defun.dk>
Message-ID: <3E624DAC.8020407@yahoo.fr>
Christian Lynbech wrote:
>>>>>>"Erik" == Erik Enge writes:
>>>>>
>
> Erik> But then I thought, where do we put these manuals once I've written
> Erik> them? We need a website to put Eclipse stuff on. What does the
> Erik> community feel about ?
>
> Sounds great. Even if there will eventually be an official eclipse
> site, there is no reason for you to not just publish it in any
> convenient way.
Sounds good to me too. Maybe we could consider to put it at the same
place the cvs repository is ? If I remember correctly it should be a
place on the machine for this purpose (like my personal account on it or
something like this). We could ask james to create us an eclipse acount
for web pages on the machine, and host the web pages there ? (who
wants to take care of them will be wellcommed)
IMHO hosting all eclipse stuff at the same place could be better.
> Once you have something to show, you should add a record at Cliki so
> other may find it.
Good point. Whatever the solution we choose.
>
> Erik> I'm writing Themers Manual right now. I'm using plain LaTeX; does
> Erik> anybody object to that?
> I would personally consider TexInfo. It is very nice for writing
> documentation and I find info pages easier to use.
>
> However, if you like LaTeX better or does not know TexInfo, you should
> keep going with LaTeX.
I agree.
!ban.
___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en fran?ais !
Yahoo! Mail : http://fr.mail.yahoo.com
From erik at nittin.net Sun Mar 2 10:03:01 2003
From: erik at nittin.net (Erik Enge)
Date: Sun Mar 2 10:03:01 2003
Subject: Eclispe website/documentation proposal.
References: <87r89qi62u.fsf@nittin.net> <87of4t1y59.fsf@baguette.defun.dk>
<3E624DAC.8020407@yahoo.fr>
Message-ID: <87llzyi04n.fsf@nittin.net>
Iban writes:
> Sounds good to me too.
And the layout and everything looks ok?
> Maybe we could consider to put it at the same place the cvs repository
> is ?
Sure, I have no problem with that.
> If I remember correctly it should be a place on the machine for this
> purpose (like my personal account on it or something like this). We
> could ask james to create us an eclipse acount for web pages on the
> machine, and host the web pages there ? (who wants to take care of
> them will be wellcommed)
I've created a script to do the nightly tarballs and I've created
Makefiles to publish the documentation whenever it is updated (this also
takes care of converting the LaTeX files into Postscript, PDF and HTML).
The only thing that would be required, I guess, would be to update the
news section every now and then and perhaps a new screenshot every so
often.
>> Once you have something to show, you should add a record at Cliki so
>> other may find it.
> Good point. Whatever the solution we choose.
Once I've got the rudimentary Manuals done I'll go ahead and update the
Cliki.
>> I would personally consider TexInfo. It is very nice for writing
>> documentation and I find info pages easier to use. However, if you
>> like LaTeX better or does not know TexInfo, you should keep going
>> with LaTeX.
> I agree.
Ok. Well, I'm not well versed in either LaTeX or TexInfo (not at all in
TexInfo, actually), but I have written pieces of the Manuals in LaTeX
already so I'll continue with that for now.
Erik.
From erik at nittin.net Sun Mar 2 10:11:01 2003
From: erik at nittin.net (Erik Enge)
Date: Sun Mar 2 10:11:01 2003
Subject: These mailinglists are now available through Gmane.
Message-ID: <87el5qhzyl.fsf@nittin.net>
I don't think I told you yet, but I took the liberty of subscribing
eclipse and eclipse-cvs to Gmane (http://www.gmane.org/).
Erik.
From hatchond at yahoo.fr Sun Mar 2 10:27:01 2003
From: hatchond at yahoo.fr (Iban)
Date: Sun Mar 2 10:27:01 2003
Subject: Eclispe website/documentation proposal.
References: <87r89qi62u.fsf@nittin.net> <87of4t1y59.fsf@baguette.defun.dk> <3E624DAC.8020407@yahoo.fr> <87llzyi04n.fsf@nittin.net>
Message-ID: <3E625A51.3020204@yahoo.fr>
Erik Enge wrote:
> Iban writes:
>
> And the layout and everything looks ok?
You mean, on the web pages you just made ?
Then yes, looks terrific.
>
>>Maybe we could consider to put it at the same place the cvs repository
>>is ?
>
>
> Sure, I have no problem with that.
>
> I've created a script to do the nightly tarballs and I've created
> Makefiles to publish the documentation whenever it is updated (this also
> takes care of converting the LaTeX files into Postscript, PDF and HTML).
> The only thing that would be required, I guess, would be to update the
> news section every now and then and perhaps a new screenshot every so
> often.
So will need accss to the cron jobs as well.
Lets decide what we need exactly and ask James about it.
> Once I've got the rudimentary Manuals done I'll go ahead and update the
> Cliki.
Ok.
> Ok. Well, I'm not well versed in either LaTeX or TexInfo (not at all in
> TexInfo, actually), but I have written pieces of the Manuals in LaTeX
> already so I'll continue with that for now.
Ok.
!ban.
___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en fran?ais !
Yahoo! Mail : http://fr.mail.yahoo.com
From erik at nittin.net Sun Mar 2 10:39:01 2003
From: erik at nittin.net (Erik Enge)
Date: Sun Mar 2 10:39:01 2003
Subject: Eclispe website/documentation proposal.
References: <87r89qi62u.fsf@nittin.net> <87of4t1y59.fsf@baguette.defun.dk>
<3E624DAC.8020407@yahoo.fr> <87llzyi04n.fsf@nittin.net>
<3E625A51.3020204@yahoo.fr>
Message-ID: <87bs0uhygd.fsf@nittin.net>
Iban writes:
> So will need accss to the cron jobs as well. Lets decide what we need
> exactly and ask James about it.
The current website comprises the following:
11 HTML pages that uses Apache's Server-Side Includes
1 cronjob that runs midnight every night
35+ files related to documentation (some images)
All in all, almost 1MB of diskspace. This will increase as the
documentation keeps growing but I doubt it will exceed 20MB. We would
need to be able to update the HTML files and the documentation
reasonably often, so shell-access is probably the best.
Erik.
From hatchond at yahoo.fr Sun Mar 2 10:45:01 2003
From: hatchond at yahoo.fr (Iban)
Date: Sun Mar 2 10:45:01 2003
Subject: Eclispe website/documentation proposal.
References: <87r89qi62u.fsf@nittin.net> <87of4t1y59.fsf@baguette.defun.dk> <3E624DAC.8020407@yahoo.fr> <87llzyi04n.fsf@nittin.net> <3E625A51.3020204@yahoo.fr> <87bs0uhygd.fsf@nittin.net>
Message-ID: <3E625E87.4020702@yahoo.fr>
Erik Enge wrote:
> Iban writes:
>
>
>>So will need accss to the cron jobs as well. Lets decide what we need
>>exactly and ask James about it.
>
>
> The current website comprises the following:
>
> 11 HTML pages that uses Apache's Server-Side Includes
> 1 cronjob that runs midnight every night
> 35+ files related to documentation (some images)
>
> All in all, almost 1MB of diskspace. This will increase as the
> documentation keeps growing but I doubt it will exceed 20MB. We would
> need to be able to update the HTML files and the documentation
> reasonably often, so shell-access is probably the best.
>
Ok. Just concerning the shell access: is ssh ok for you ? (it is the
current way to acces on the unlambda machine)
___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en fran?ais !
Yahoo! Mail : http://fr.mail.yahoo.com
From erik at nittin.net Sun Mar 2 10:47:01 2003
From: erik at nittin.net (Erik Enge)
Date: Sun Mar 2 10:47:01 2003
Subject: Eclispe website/documentation proposal.
References: <87r89qi62u.fsf@nittin.net> <87of4t1y59.fsf@baguette.defun.dk>
<3E624DAC.8020407@yahoo.fr> <87llzyi04n.fsf@nittin.net>
<3E625A51.3020204@yahoo.fr> <87bs0uhygd.fsf@nittin.net>
<3E625E87.4020702@yahoo.fr>
Message-ID: <878yvyhy3n.fsf@nittin.net>
Iban writes:
> Ok. Just concerning the shell access: is ssh ok for you ? (it is the
> current way to acces on the unlambda machine)
SSH is what I use myself so that will be just fine.
Erik.
From erik at nittin.net Sun Mar 2 11:14:00 2003
From: erik at nittin.net (Erik Enge)
Date: Sun Mar 2 11:14:00 2003
Subject: Try loading ~/.eclipse/custom.lisp before ~/.eclipse
References: <873cm7jlb5.fsf@nittin.net> <87r89p1ykz.fsf@baguette.defun.dk>
Message-ID: <8765r2hwut.fsf@nittin.net>
Christian Lynbech writes:
> How about making the default config file name be a configuration
> option and perhaps adding a switch to the `eclipse' script that allows
> a per-session specification to be made?
Also, since the configuration is loaded before we load the initial
theme, how about letting the user specify the initial theme in the
~/.eclipse too? I'm unsure about the best approach, but perhaps we
could do:
(setf (decoration-theme) *inital-theme*))
(init-edges-cursors)
(initialize-clone))))
instead of:
(setf (decoration-theme) "microGUI"))
(init-edges-cursors)
(initialize-clone))))
And then, in globals.lisp, we'd say:
(defparameter *initial-theme* "microGUI")
Thus, if the user wants brushed-metal as his default theme, he could put
the following in his ~/.eclipse:
(defparameter *initial-theme* "brushed-metal")
Erik.
From erik at nittin.net Sun Mar 2 14:41:01 2003
From: erik at nittin.net (Erik Enge)
Date: Sun Mar 2 14:41:01 2003
Subject: %QUIT% called one time too many?
Message-ID: <87znodhn91.fsf@nittin.net>
In package.lisp, we have:
(defun start-eclipse (display)
(eclipse:eclipse display)
(eclipse::%quit%))
ECLIPSE:ECLIPSE (which resides in eclipse.lisp) will also call %QUIT%
before returning; doesn't that make the %QUIT% here redundant?
Erik.
From hatchond at yahoo.fr Sun Mar 2 14:50:01 2003
From: hatchond at yahoo.fr (Iban)
Date: Sun Mar 2 14:50:01 2003
Subject: %QUIT% called one time too many?
References: <87znodhn91.fsf@nittin.net>
Message-ID: <3E62981E.3010307@yahoo.fr>
Erik Enge wrote:
> In package.lisp, we have:
>
> (defun start-eclipse (display)
> (eclipse:eclipse display)
> (eclipse::%quit%))
>
> ECLIPSE:ECLIPSE (which resides in eclipse.lisp) will also call %QUIT%
> before returning; doesn't that make the %QUIT% here redundant?
Probably. I don't remember what I got in mind when I wrote that, but it
looks useless.
___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en fran?ais !
Yahoo! Mail : http://fr.mail.yahoo.com
From erik at nittin.net Sun Mar 2 18:16:01 2003
From: erik at nittin.net (Erik Enge)
Date: Sun Mar 2 18:16:01 2003
Subject: Dynamic binding of keyname/keysym. [oops, sent the previous one
too fast]
In-Reply-To: <1046459661.3e5fb50d8ac1f@iona.labri.fr> (hatchond@labri.fr's
message of "Fri, 28 Feb 2003 20:14:21 +0100")
References: <3E5F9C76.2080805@nittin.net>
<1046459661.3e5fb50d8ac1f@iona.labri.fr>
Message-ID: <87vfz1hdc1.fsf_-_@nittin.net>
hatchond at labri.fr writes:
> Should be normal. Because we don't handle the :mapping-notify event (clx
> 12.12.3), but we should handle it. If you look at the definition of this event,
> you'll see that we have a chance to handle this correctly.
> But first, I would like to know what happens if you change the keyboard map
> before lunching eclipse ? I don't figure how to handle a :mapping-notify with
> request :keyboard correctly for the moment, maybe you do.
> I was waiting for somebody like you (with strange keyboard mapping) to point
> this out. (I wasn't sure of the necessity of handling this event before :o( )
[snip]
> The trouble is: I don't have the (even the begining of) solution in my
> mind for the moment
According to the CLX (12.12.3), if REQUEST is :KEYBOARD then START and
COUNT indicate the range of the altered keycodes.
A snip from my ~/.Xmodmap file:
keycode 100 = Left
keycode 101 =
keycode 102 = Right
keycode 103 = End
keycode 104 = Down
keycode 105 = Next
keycode 106 = Insert
keycode 107 = Delete
keycode 108 = KP_Enter
keycode 109 = Control_R
keycode 110 = Pause Break
keycode 111 = Print Execute
When I run `xmodmap ~/.Xmodmap` Eclipse receives one MAPPING-NOTIFY
event for each line in the .Xmodmap file. The 'keycode' in the .Xmodmap
file ends up in the START slot of the MAPPING-NOTIFY event.
Now, seeing as the DEFINE-KEYSYM function takes `name` and `value`, I
assume I should be able to do something like this:
(define-keysym (keycode->keyname *display* (event-start event))
[assuming 'event' is a MAPPING-NOTIFY event object]
But that doesn't seem to make any sense as KEYCODE->KEYNAME seems to
want hexadecimal values for the keycodes.
Am I on the right track here? Should I just convert the 'keycode' to
hexadecimal and that's it? If so, we should perhaps have a function
that's called UPDATE-KEYSYM (or somesuch) as that would give more
meaning (at least to me).
Erik.
From erik at nittin.net Sun Mar 2 20:48:00 2003
From: erik at nittin.net (Erik Enge)
Date: Sun Mar 2 20:48:00 2003
Subject: Dynamic binding of keyname/keysym.
In-Reply-To: <1046459661.3e5fb50d8ac1f@iona.labri.fr> (hatchond@labri.fr's
message of "Fri, 28 Feb 2003 20:14:21 +0100")
References: <3E5F9C76.2080805@nittin.net>
<1046459661.3e5fb50d8ac1f@iona.labri.fr>
Message-ID: <87wujhhdgf.fsf@nittin.net>
hatchond at labri.fr writes:
> Should be normal. Because we don't handle the :mapping-notify event (clx
> 12.12.3), but we should handle it. If you look at the definition of this event,
> you'll see that we have a chance to handle this correctly.
> But first, I would like to know what happens if you change the keyboard map
> before lunching eclipse ? I don't figure how to handle a :mapping-notify with
> request :keyboard correctly for the moment, maybe you do.
> I was waiting for somebody like you (with strange keyboard mapping) to point
> this out. (I wasn't sure of the necessity of handling this event before :o( )
[snip]
> The trouble is: I don't have the (even the begining of) solution in my
> mind for the moment
According to the CLX (12.12.3), if REQUEST is :KEYBOARD then START and
COUNT indicate the range of the altered keycodes.
A snip from my ~/.Xmodmap file:
keycode 100 = Left
keycode 101 =
keycode 102 = Right
keycode 103 = End
keycode 104 = Down
keycode 105 = Next
keycode 106 = Insert
keycode 107 = Delete
keycode 108 = KP_Enter
keycode 109 = Control_R
keycode 110 = Pause Break
keycode 111 = Print Execute
When I run `xmodmap ~/.Xmodmap` Eclipse receives one MAPPING-NOTIFY
event for each line in the .Xmodmap file. The 'keycode' in the .Xmodmap
file ends up in the START slot of the MAPPING-NOTIFY event.
Now, seeing as the DEFINE-KEYSYM function takes `name` and `value`, I
assume I should be able to do something like this:
(define-keysym (keycode->keyname *display* (event-start event))
(defun define-keysym (name value)
(pushnew name (gethash value *keysym->keyname* nil))
(setf (gethash name *keyname->keysyms*) value))
(defun keysym->keyname (keysym)
(car (last (gethash keysym *keysym->keyname*))))
(defun keyname->keysym (keyname)
(gethash keyname *keyname->keysyms*))
(defun keyname->keycodes (disp keyname)
(xlib:keysym->keycodes disp (keyname->keysym keyname)))
(defun keycode->keyname (disp keycode)
(keysym->keyname (xlib:keycode->keysym disp keycode 0)))
From strandh at labri.u-bordeaux.fr Mon Mar 3 00:51:01 2003
From: strandh at labri.u-bordeaux.fr (Robert STRANDH)
Date: Mon Mar 3 00:51:01 2003
Subject: Eclispe website/documentation proposal.
In-Reply-To: <87r89qi62u.fsf@nittin.net>
References: <87r89qi62u.fsf@nittin.net>
Message-ID: <15970.63530.232334.328428@serveur3.labri.fr>
Erik Enge writes:
>
> I'm writing Themers Manual right now. I'm using plain LaTeX; does
> anybody object to that?
I think that's an excellent idea. I have heard people in the past
voting for XML instead, but it seems like a bad idea to write XML
directly, mainly because there is no macro facility.
--
Robert Strandh
---------------------------------------------------------------------
Greenspun's Tenth Rule of Programming: any sufficiently complicated C
or Fortran program contains an ad hoc informally-specified bug-ridden
slow implementation of half of Common Lisp.
---------------------------------------------------------------------
From strandh at labri.u-bordeaux.fr Mon Mar 3 00:51:03 2003
From: strandh at labri.u-bordeaux.fr (Robert STRANDH)
Date: Mon Mar 3 00:51:03 2003
Subject: Eclispe website/documentation proposal.
In-Reply-To: <87of4t1y59.fsf@baguette.defun.dk>
References: <87r89qi62u.fsf@nittin.net>
<87of4t1y59.fsf@baguette.defun.dk>
Message-ID: <15970.63664.251131.336514@serveur3.labri.fr>
Christian Lynbech writes:
> Erik> I'm writing Themers Manual right now. I'm using plain LaTeX; does
> Erik> anybody object to that?
>
> I would personally consider TexInfo. It is very nice for writing
> documentation and I find info pages easier to use.
Yes, TexInfo would be fine too.
--
Robert Strandh
---------------------------------------------------------------------
Greenspun's Tenth Rule of Programming: any sufficiently complicated C
or Fortran program contains an ad hoc informally-specified bug-ridden
slow implementation of half of Common Lisp.
---------------------------------------------------------------------
From erik at nittin.net Sat Mar 15 11:23:01 2003
From: erik at nittin.net (Erik Enge)
Date: Sat Mar 15 11:23:01 2003
Subject: Differences between event.lisp and lib/clx-ext/event.lisp
Message-ID: <878yvgzbbu.fsf@nittin.net>
The two files are quite similar (8 and 9k respectively). They contain
alot of the same code and I don't understand why this is so. Is there
any reason for this that I am missing?
I was quite confused when i TRACEd ./event.lisp:clos-event-handler and
nothing showed up so unless there is a reason for this we should
probably clean it up.
Erik.
From hatchond at yahoo.fr Sat Mar 15 12:08:00 2003
From: hatchond at yahoo.fr (=?iso-8859-1?q?Iban=20HATCHONDO?=)
Date: Sat Mar 15 12:08:00 2003
Subject: Differences between event.lisp and lib/clx-ext/event.lisp
In-Reply-To: <878yvgzbbu.fsf@nittin.net>
Message-ID: <20030315200717.1760.qmail@web40614.mail.yahoo.com>
--- Erik Enge a ?crit : > The two
files are quite similar (8 and 9k
> respectively). They contain
> alot of the same code and I don't understand why
> this is so. Is there
> any reason for this that I am missing?
>
> I was quite confused when i TRACEd
> ./event.lisp:clos-event-handler and
> nothing showed up so unless there is a reason for
> this we should
> probably clean it up.
>
> Erik.
My fault. This file has moved in lib/clx-ext and I
fogot to removed it from its first place. I'll do that
soon.
Thanks for pointing this.
!ban.
___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en fran?ais !
Yahoo! Mail : http://fr.mail.yahoo.com
From eenge at prium.net Wed Mar 19 06:45:01 2003
From: eenge at prium.net (Erik Enge)
Date: Wed Mar 19 06:45:01 2003
Subject: Comments on focusing (on new code in CVS)
Message-ID: <877kav30sc.fsf@prium.net>
I've done a checkout of the code in CVS. It works well, but I have a
couple of more comments with regards to the focusing, under `:on-click`:
- If I have two windows. Window 1 is focused and window 2 is behind
window 1 but I can see both. If I now hit the maximize button on
window 2, I'd expect it to become focused as well as maximized.
Currently, it only gets maximized.
- The same behaviour exist for the menu button, it does not focus the
window 1 (if window 2 is focused) when I click its menu button.
Other than that, I haven't been able to find anything to complain about.
If I get some time this week, I'll see if I can fix both of the above
issues (assuming you all agree with me that they are issues that needs
to be fixed).
Erik.
From lynbech at defun.dk Sun Mar 23 10:28:01 2003
From: lynbech at defun.dk (Christian Lynbech)
Date: Sun Mar 23 10:28:01 2003
Subject: patches for eclipse building
Message-ID: <15997.60344.849921.612164@baguette.defun.dk>
I have finally finished my attempt at improving the eclipse build
process.
As far as I can test it, it should support both fasl files (libing in the
source directory) and standalone core files (where everything
necessary gets installed).
This patch contains also a new option for determining the name of the
startup file as was previously discussed.
I propose (and the patch anticipates its approval :-) that we rename
the system startup file from .eclipse to eclipserc. I do not think
eclipse should install dot files. I always keep forgetting that theyu
are there and they is no real reason that the two files (user file and
library file) must have the same name.
I have thrown in another little modification. Under CMUCL, I have
changed the %quit% code from calling `quit' to calling `unix:exit'. I
have numerous times gotten an eclipse process looping endslessly in a
SIGPIPE error after exiting X. I guess that `quit' wants to do some
cleanup, but this fails since the connection has been closed
down. After the change I have not seen that problem. Alternatively we
could probably just call `quit' with a non-nil argument which
presumably has pretty much the same effect (judging from the document
string of `quit').
I have tried to add some documentation to the readme file but I am not
sure that I am explaining configuration options very well.
The patch comes as three files, one diff, the new file
"config.lisp.in" and a cvsignore file for lib/ which I think is
missing.
The diff is against a freshly updated sandbox so it should apply
cleanly.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: diff2
Type: application/octet-stream
Size: 25414 bytes
Desc: not available
URL:
-------------- next part --------------
A non-text attachment was scrubbed...
Name: config.lisp.in
Type: application/octet-stream
Size: 1086 bytes
Desc: lisp configuration helper file
URL:
-------------- next part --------------
A non-text attachment was scrubbed...
Name: .cvsignore
Type: application/octet-stream
Size: 6 bytes
Desc: cvsignore file for lib/
URL:
-------------- next part --------------
------------------------+-----------------------------------------------------
Christian Lynbech | email: christian at defun.dk
------------------------+-----------------------------------------------------
Hit the philistines three times over the head with the Elisp reference manual.
- petonic at hal.com (Michael A. Petonic)
From hatchond at labri.fr Sun Mar 23 13:52:01 2003
From: hatchond at labri.fr (hatchond at labri.fr)
Date: Sun Mar 23 13:52:01 2003
Subject: patches for eclipse building
In-Reply-To: <15997.60344.849921.612164@baguette.defun.dk>
References: <15997.60344.849921.612164@baguette.defun.dk>
Message-ID: <1048459901.3e7e3a7da5c83@iona.labri.fr>
I'll try to look and play with this on monday. I think I can commit monday night
(here) or thuesday. If anyone wants to add or comments, please go ahead.
Iban.
-------------------------------------------------
This mail sent through IMP: http://horde.org/imp/
From haalst at online.no Sun Mar 23 20:32:00 2003
From: haalst at online.no (Håkon Alstadheim)
Date: Sun Mar 23 20:32:00 2003
Subject: patches for eclipse building
In-Reply-To: <15997.60344.849921.612164@baguette.defun.dk>
Message-ID: <200303232320.h2NNKYD8014199@alstadhome.dyndns.org>
Christian Lynbech writes:
> I have thrown in another little modification. Under CMUCL, I have
> changed the %quit% code from calling `quit' to calling `unix:exit'. I
> have numerous times gotten an eclipse process looping endslessly in a
> SIGPIPE error after exiting X. I guess that `quit' wants to do some
> cleanup, but this fails since the connection has been closed
> down. After the change I have not seen that problem. Alternatively we
> could probably just call `quit' with a non-nil argument which
> presumably has pretty much the same effect (judging from the document
> string of `quit').
CMUCL 18e will have a QUIT that only takes keyword parameters. Better
to use UNIX:EXIT.
From hatchond at yahoo.fr Mon Mar 24 05:00:02 2003
From: hatchond at yahoo.fr (=?iso-8859-1?q?Iban=20HATCHONDO?=)
Date: Mon Mar 24 05:00:02 2003
Subject: patches for eclipse building
In-Reply-To: <200303232320.h2NNKYD8014199@alstadhome.dyndns.org>
Message-ID: <20030324135929.8517.qmail@web40605.mail.yahoo.com>
--- H?kon Alstadheim a ?crit : >
>
> CMUCL 18e will have a QUIT that only takes keyword
> parameters. Better
> to use UNIX:EXIT.
Ok. So lets do it as Christian made it.
By the way the way you may have some suggestion on a
problem I got with cmucl and compile-file-pathname:
on a 18d release on my machine:
* (compile-file-pathname "themes/microGUI/theme.lisp"
:output-file "theme.o")
#p"theme.o"
*
and on Erik's machine (as well as Christian's one)
the same gives #p"themes/microGUI/theme.o" ?
Any clues ?
Iban.
___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en fran?ais !
Yahoo! Mail : http://fr.mail.yahoo.com
From hatchond at labri.fr Mon Mar 24 08:45:02 2003
From: hatchond at labri.fr (Iban Hatchondo)
Date: Mon Mar 24 08:45:02 2003
Subject: patches for eclipse building
References: <15997.60344.849921.612164@baguette.defun.dk>
Message-ID: <3E7F4409.2070206@labri.fr>
Christian Lynbech wrote:
> I have finally finished my attempt at improving the eclipse build
> process.
>
- In Makefile.in, in the install clause:
if test ! -x ${bindir}/`basename ${LISP}` ; then \
$(install) ${LISP} ${bindir} ; \
fi
what will happend if user did not dump an eclipse lisp image, and he
upgrade his lisp system ?
I suspect he will got a copy of his old lisp trying to load an old lisp
image, what should failed ?
What could we do to avoid this ? Not use this method in that case ?
(this implie modification in eclipse.in)
- I want to clarify something about the config.lisp file you add:
if I try to compile eclipse after applying your diff I got the
followings warnings. I know they are not a problem considering that the
undefined variables will be defined during the lisp image dumping, or
during the starting process with load-eclipse.
The question is: why didn't you add this file in the eclipse-system
defined in system.lisp ? That way, we shouldn't have compilation
warnings, and load-eclipse should not anymore have to load that file
explicitly.
I presume I missed something ?
!ban.
-- part of my compilation output. -- (cmucl 18e-pre1 for testing)
[ snip ]
; Python version 1.1, VM version Intel x86 on 24 MAR 03 05:15:44 pm.
; Compiling: /tmp/eclipse/global.lisp 24 MAR 03 05:02:17 pm
;
;
; File: /tmp/eclipse/global.lisp
; In: DEFUN ECLIPSE-PATH
; (OR COMMON-LISP-USER::*ECLIPSE-ECLIPSEDIR* *ECLIPSE-DIRECTORY*)
; ==>
; (LET (#)
; (IF #:G2 #:G2 #))
; Warning: Undefined variable: COMMON-LISP-USER::*ECLIPSE-ECLIPSEDIR*
;
; In: LAMBDA (&REST NAMES)
; (OR COMMON-LISP-USER::*ECLIPSE-ECLIPSEDIR* *ECLIPSE-DIRECTORY*)
; ==>
; (LET (#)
; (IF #:G0 #:G0 #))
; Warning: Undefined variable: COMMON-LISP-USER::*ECLIPSE-ECLIPSEDIR*
; ;
; Warning: This variable is undefined:
; COMMON-LISP-USER::*ECLIPSE-ECLIPSEDIR*
;
; Compilation unit finished.
; 3 warnings
; /tmp/eclipse/global.x86f written.
; Compilation finished in 0:00:01.
;; Loading #p"/tmp/eclipse/global.x86f".
[ snip ]
; Python version 1.1, VM version Intel x86 on 24 MAR 03 05:16:01 pm.
; Compiling: /tmp/eclipse/eclipse.lisp 24 MAR 03 05:02:17 pm
;
;
; File: /tmp/eclipse/eclipse.lisp
; In: DEFUN INITIALIZE
; (HOME-SUBDIRECTORY COMMON-LISP-USER::*ECLIPSE-RCFILE*)
; Warning: Undefined variable: COMMON-LISP-USER::*ECLIPSE-RCFILE*
;
; In: LAMBDA (DISPLAY-SPECIFICATION)
; (HOME-SUBDIRECTORY COMMON-LISP-USER::*ECLIPSE-RCFILE*)
; Warning: Undefined variable: COMMON-LISP-USER::*ECLIPSE-RCFILE*
; ;
; Warning: This variable is undefined:
; COMMON-LISP-USER::*ECLIPSE-RCFILE*
;
; Compilation unit finished.
; 3 warnings
; /tmp/eclipse/eclipse.x86f written.
; Compilation finished in 0:00:00.
;; Loading #p"/tmp/eclipse/eclipse.x86f".
From christian at defun.dk Mon Mar 24 12:04:01 2003
From: christian at defun.dk (Christian Lynbech)
Date: Mon Mar 24 12:04:01 2003
Subject: patches for eclipse building
In-Reply-To: <3E7F4409.2070206@labri.fr> (Iban Hatchondo's message of "Mon,
24 Mar 2003 18:44:41 +0100")
References: <15997.60344.849921.612164@baguette.defun.dk>
<3E7F4409.2070206@labri.fr>
Message-ID: <873clczdcz.fsf@baguette.defun.dk>
>>>>> "Iban" == Iban Hatchondo writes:
Iban> - In Makefile.in, in the install clause:
Iban> if test ! -x ${bindir}/`basename ${LISP}` ; then \
Iban> $(install) ${LISP} ${bindir} ; \
Iban> fi
Iban> what will happend if user did not dump an eclipse lisp image, and he
Iban> upgrade his lisp system ?
You are right. I think we should move that clause up into the
${usecore} if-block above such that saving of lisp program only is
active when `usecore' is active.
Iban> The question is: why didn't you add this file in the eclipse-system
Iban> defined in system.lisp ? That way, we shouldn't have compilation
Iban> warnings, and load-eclipse should not anymore have to load that file
Iban> explicitly.
An oversight on my part, I think. There are so many warnings that I am
not trying to monitor them closely, but this we should fix especially
since it seems easy.
------------------------+-----------------------------------------------------
Christian Lynbech | email: christian at defun.dk
------------------------+-----------------------------------------------------
Hit the philistines three times over the head with the Elisp reference manual.
- petonic at hal.com (Michael A. Petonic)
From hatchond at yahoo.fr Mon Mar 24 12:59:01 2003
From: hatchond at yahoo.fr (=?iso-8859-1?q?Iban=20HATCHONDO?=)
Date: Mon Mar 24 12:59:01 2003
Subject: patches for eclipse building
In-Reply-To: <873clczdcz.fsf@baguette.defun.dk>
Message-ID: <20030324215747.74063.qmail@web40603.mail.yahoo.com>
--- Christian Lynbech a ?crit :
>
> Iban> what will happend if user did not dump an
> eclipse lisp image, and he
> Iban> upgrade his lisp system ?
>
> You are right. I think we should move that clause up
> into the
> ${usecore} if-block above such that saving of lisp
> program only is
> active when `usecore' is active.
Ok. I'll try to fix it as well with your patch.
> An oversight on my part, I think. There are so many
> warnings that I am
> not trying to monitor them closely, but this we
> should fix especially
> since it seems easy.
Ok. I'll fix it as well.
But I am suprise you say you got as much warnings. It
should only be four files that gives warnings:
- lib/clx-ext/clx-patch.err which doesn't matter.
- lib/image-reader.err lib/manager-commons.err: 7
- input.lisp: 2
you guies got more than that ?
I am querying because warnings are potentials bugs,
and so I'd like to fix them as well as we could. (I
never found how to fix the ones I mension up there)
!ban.
___________________________________________________________
Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en fran?ais !
Yahoo! Mail : http://fr.mail.yahoo.com
From james at unlambda.com Mon Mar 24 14:19:00 2003
From: james at unlambda.com (James A. Crippen)
Date: Mon Mar 24 14:19:00 2003
Subject: Eclispe website/documentation proposal.
In-Reply-To: <878yvyhy3n.fsf@nittin.net> (Erik Enge's message of "Sun, 02
Mar 2003 05:17:00 -0500")
References: <87r89qi62u.fsf@nittin.net> <87of4t1y59.fsf@baguette.defun.dk>
<3E624DAC.8020407@yahoo.fr> <87llzyi04n.fsf@nittin.net>
<3E625A51.3020204@yahoo.fr> <87bs0uhygd.fsf@nittin.net>
<3E625E87.4020702@yahoo.fr> <878yvyhy3n.fsf@nittin.net>
Message-ID:
Erik Enge writes:
> Iban writes:
>
>> Ok. Just concerning the shell access: is ssh ok for you ? (it is the
>> current way to acces on the unlambda machine)
>
> SSH is what I use myself so that will be just fine.
>
> Erik.
Mail me and I can set up a shared account for the Eclipse people.
It'll have ssh, web, cvs, etc.
Sorry for not having paid attention lately, my life's been a bit busy
lately.
'james
--
James A. Crippen Lambda Unlimited
61.2204N, -149.8964W Recursion 'R' Us
Anchorage, Alaska, USA, Earth Y = \f.(\x.f(xx))(\x.f(xx))
From erik at nittin.net Mon Mar 24 18:30:01 2003
From: erik at nittin.net (Erik Enge)
Date: Mon Mar 24 18:30:01 2003
Subject: Comments on focusing (on new code in CVS)
References: <877kav30sc.fsf@prium.net>
Message-ID: <87of40nn28.fsf@nittin.net>
Erik Enge writes:
> - If I have two windows. Window 1 is focused and window 2 is behind
> window 1 but I can see both. If I now hit the maximize button on
> window 2, I'd expect it to become focused as well as maximized.
> Currently, it only gets maximized.
I'm trying to fix that with the following code in input.lisp:
;; Maximization
(defmethod event-process ((event button-release) (max-b maximize-button))
(when (< (event-code event) 4)
(with-slots (master window) max-b
(when (eq *focus-type* :on-click)
(vs:restack-window window (vs:nth-vscreen (root-vscreens *root*))))
(when master (draw-focused-decoration master))
(setf (netwm:net-active-window *root-window*) window))
(maximize-window (button-master max-b) (event-code event))))
It used to look like this:
;; Maximization
(defmethod event-process ((event button-release) (max-b maximize-button))
(when (< (event-code event) 4)
(maximize-window (button-master max-b) (event-code event))))
As the attached screenshot shows, the window is maximized and partly
focused. Two problems exist:
"partly focused" - as you can see (screenshot) the bigger window (the
one just maximized) is not focused (or has not loaded the pixmap that
represents the focusing) the title-bar. I can't explain why.
The window that was previously focus is, well, still focused. :) Do I
need to call some more code to defocus the current window before
focusing a new one, perhaps?
Thanks for any help,
Erik.
From hatchond at labri.fr Tue Mar 25 00:36:02 2003
From: hatchond at labri.fr (Iban Hatchondo)
Date: Tue Mar 25 00:36:02 2003
Subject: Comments on focusing (on new code in CVS)
References: <877kav30sc.fsf@prium.net> <87of40nn28.fsf@nittin.net>
Message-ID: <3E8022BD.8060601@labri.fr>
Erik Enge wrote:
> Erik Enge writes:
>
>
>> - If I have two windows. Window 1 is focused and window 2 is behind
>> window 1 but I can see both. If I now hit the maximize button on
>> window 2, I'd expect it to become focused as well as maximized.
>> Currently, it only gets maximized.
>
>
> I'm trying to fix that with the following code in input.lisp:
>
> ;; Maximization
> (defmethod event-process ((event button-release) (max-b maximize-button))
> (when (< (event-code event) 4)
> (with-slots (master window) max-b
> (when (eq *focus-type* :on-click)
> (vs:restack-window window (vs:nth-vscreen (root-vscreens *root*))))
> (when master (draw-focused-decoration master))
> (setf (netwm:net-active-window *root-window*) window))
> (maximize-window (button-master max-b) (event-code event))))
>
[ snip ]
>
> As the attached screenshot shows, the window is maximized and partly
> focused. Two problems exist:
I can't see any attached files.
> "partly focused" - as you can see (screenshot) the bigger window (the
> one just maximized) is not focused (or has not loaded the pixmap that
> represents the focusing) the title-bar. I can't explain why.
>
> The window that was previously focus is, well, still focused. :) Do I
> need to call some more code to defocus the current window before
> focusing a new one, perhaps?
>
> Thanks for any help,
I think you went the wrong way:
- if you use draw-focused-decoration, the decoration should be focused
what is not the fact here.
Why don't you just try to call (focus-widget max-b 0) ?
It will gives the focus to the decoration according to focus rules (in
wm.lisp) and the X events will do the rest: now the application got
focus (according focus rules), a focus-in event will be generate on our
display. And so, the event-process method for focus-in will be called.
This method will restack the window if needed, draw focused decoration
if decorated, and set the net-active-window property.
(defmethod event-process ((event button-release) (max-b maximize-button))
(when (< (event-code event) 4)
(maximize-window (button-master max-b) (event-code event))
(focus-widget max-b 0)))
!ban.
From erik at nittin.net Tue Mar 25 13:51:02 2003
From: erik at nittin.net (Erik Enge)
Date: Tue Mar 25 13:51:02 2003
Subject: Comments on focusing (on new code in CVS)
References: <877kav30sc.fsf@prium.net> <87of40nn28.fsf@nittin.net>
<3E8022BD.8060601@labri.fr>
Message-ID: <87isu7ysdp.fsf@nittin.net>
Iban Hatchondo writes:
> I can't see any attached files.
D'oh. Anyway, it seems that it was fixed when I used focus-widget
> Why don't you just try to call (focus-widget max-b 0)?
Didn't know about it. But I do now!
Here's the patch I'd like see included in input.lisp:
*** /tmp/erik/input.lisp 2003-03-25 17:47:01.000000000 -0500
--- /tmp/erik/input.lispUEA3HV 2003-03-25 17:47:01.000000000 -0500
***************
*** 389,394 ****
--- 389,395 ----
(defmethod event-process ((event button-press) (button menu-button))
(with-slots (window-menu) *root*
(with-slots (master window armed active-p) button
+ (focus-widget button 0)
(and window-menu (destroy-substructure window-menu))
(setf window-menu (make-menu-button-menu master))
(realize-pop-up window-menu (event-root-x event) (event-root-y event))
***************
*** 397,403 ****
;; Maximization
(defmethod event-process ((event button-release) (max-b maximize-button))
(when (< (event-code event) 4)
! (maximize-window (button-master max-b) (event-code event))))
;; Initialize the resize process.
(defmethod event-process ((event button-press) (edge edge))
--- 398,405 ----
;; Maximization
(defmethod event-process ((event button-release) (max-b maximize-button))
(when (< (event-code event) 4)
! (maximize-window (button-master max-b) (event-code event))
! (focus-widget max-b 0)))
;; Initialize the resize process.
(defmethod event-process ((event button-press) (edge edge))
It will focus the window when you click the menu-button and the
maximization-button. I'm not testing for :on-click since I guess that
you'd get focus just by moving your mouse over the buttons anyway.
Erik.
From erik at nittin.net Tue Mar 25 17:54:01 2003
From: erik at nittin.net (Erik Enge)
Date: Tue Mar 25 17:54:01 2003
Subject: Problem with focusing (:on-click) when changing vscreens.
Message-ID: <87fzpa2634.fsf@nittin.net>
Using the latest CVS version:
Say I have four virtual screens. On screen one I have none or several
windows (doesn't seem to affect the behaviour). On screen two I have an
xterm running (or Mozilla, or Emacs, again, this doesn't seem to
matter). If I switch from virtual-screen one to virtual-screen two[1]
the xterm gets the focus every time I enter virtual-screen two. This is
as expected.
Now, if I am on virtual-screen two and start a new xterm[2] and then
switch to virtual-screen one using my keycombo /without/ clicking in the
xterm window the problem shows upon returning to virtual-screen two.
The new xterm (which should be focused, because it was the one that was
focused when I left virtual-screen two) is not focused, it is the first
xterm that has the focus.
The problem goes away if I click with my mouse in the newly created
xterm window before leaving virtual-screen two; upon returning, this
xterm is now focused and all is as expected.
The solution might lay in `change-vscreen` (widgets.lisp) but I'm unable
to see it. Perhaps someone else would be able to shed some light on
this?
I guess this would only show up under :on-click, which is what I am
using.
Erik.
Footnotes:
[1] I use a custom keystroke:
;; Circulate virtual screens
(define-key-combo :switch-screen-left
:keys '(:p)
:modifiers '(:and :alt-left :control-left))
(define-key-combo :switch-screen-right
:keys '(:n)
:modifiers '(:and :alt-left :control-left))
[2] I use a custom keystroke for this too:
;; Application shortcuts
(define-key-combo :start-xterm
:keys '(:x)
:modifiers '(:and :alt-left :control-left)
:fun (action (:press (%run-program% "xterm" '())) ()))
From hatchond at labri.fr Wed Mar 26 06:44:00 2003
From: hatchond at labri.fr (Iban Hatchondo)
Date: Wed Mar 26 06:44:00 2003
Subject: Problem with focusing (:on-click) when changing vscreens.
References: <87fzpa2634.fsf@nittin.net>
Message-ID: <3E81CA84.8090006@labri.fr>
Erik Enge wrote:
[ ... ]
> Using the latest CVS version:
> The problem goes away if I click with my mouse in the newly created
> xterm window before leaving virtual-screen two; upon returning, this
> xterm is now focused and all is as expected.
>
> The solution might lay in `change-vscreen` (widgets.lisp) but I'm unable
> to see it. Perhaps someone else would be able to shed some light on
> this?
It seems the problem comes as you suspect from the keystroke :statr-xterm.
Apparently, it interfere with the focus-in event we receive next.
As you said we could fix this by calling restack-window in
change-vscreen, but it
may point out a deeper problem. So I would try another solution, but I
am not completely sure. Could try to replace the
(event-process ((event keyboard-event) (root root)) ..) in input.lisp by
(defmethod event-process ((event keyboard-event) (root root))
(with-slots (code state) event
(let ((callback (lookup-keystroke code state)))
(when callback
(xlib:allow-events *display* :async-keyboard)
(funcall callback event)))))
Note the allow-events that now appear. It should fix our problem, but
I think it should also be more correct that way (in case this interfere
with more than what we though).
Iban.
> Footnotes:
> [2] I use a custom keystroke for this too:
>
> ;; Application shortcuts
>
> (define-key-combo :start-xterm
> :keys '(:x)
> :modifiers '(:and :alt-left :control-left)
> :fun (action (:press (%run-program% "xterm" '())) ()))
>
(defmethod event-process ((event keyboard-event) (root root))
(with-slots (code state) event
(let ((callback (lookup-keystroke code state)))
(when callback
(xlib:allow-events *display* :async-keyboard)
(funcall callback event)))))
From erik at nittin.net Wed Mar 26 16:26:01 2003
From: erik at nittin.net (Erik Enge)
Date: Wed Mar 26 16:26:01 2003
Subject: Problem with focusing (:on-click) when changing vscreens.
References: <87fzpa2634.fsf@nittin.net> <3E81CA84.8090006@labri.fr>
Message-ID: <8765q5a9ha.fsf@nittin.net>
Iban Hatchondo writes:
> So I would try another solution, but I am not completely sure. Could
> try to replace the (event-process ((event keyboard-event) (root root))
> ..) in input.lisp by [snip]
Yep, cool, this seems to fix the problem. Thanks! :)
Erik.