Menu
▾
▴
libvisual-devel — libvisual mailinglist for developers
You can subscribe to this list here.
| 2004 |
Jan
|
Feb
|
Mar
|
Apr
(38) |
May
(22) |
Jun
(92) |
Jul
(101) |
Aug
(18) |
Sep
(286) |
Oct
(180) |
Nov
(73) |
Dec
(14) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2005 |
Jan
(18) |
Feb
(74) |
Mar
(56) |
Apr
(11) |
May
(5) |
Jun
(4) |
Jul
(20) |
Aug
(4) |
Sep
|
Oct
|
Nov
(1) |
Dec
(2) |
| 2006 |
Jan
(11) |
Feb
(2) |
Mar
(10) |
Apr
(2) |
May
(1) |
Jun
|
Jul
(24) |
Aug
(11) |
Sep
(5) |
Oct
(16) |
Nov
(25) |
Dec
(8) |
| 2007 |
Jan
|
Feb
(1) |
Mar
|
Apr
|
May
|
Jun
(3) |
Jul
(1) |
Aug
|
Sep
|
Oct
(4) |
Nov
(12) |
Dec
|
| 2009 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
(3) |
Aug
|
Sep
|
Oct
|
Nov
(2) |
Dec
|
| 2010 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
(1) |
Dec
|
| 2011 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
(5) |
|
From: Duilio J. P. <dp...@fc...> - 2004-11-24 23:27:46
|
I have found a way to fix this, the only issue is that glib will doesn't have knowledge about the private info, but I think that's not a problem on our context (the destroyer will get data he needs). I hope to have the patch tomorrow. Bye, Duilio. > Suggestions ?, we could have an extra entry in the object instance > structure.. > > Cheers, > Dennis |
|
From: Dennis S. <sy...@yo...> - 2004-11-24 20:29:31
|
On Wed, 2004-11-24 at 19:29 +0200, Vitaly V. Bursov wrote: > On Tue, 23 Nov 2004 19:13:44 +0100 > Dennis Smit <sy...@yo...> wrote: > > > Vitaly, I can't use the GLX drivers because it currently only supports > > GLX 1.3, and the nvidia drivers I have are 1.2... What is the actual > That's wrong. I know no nvidia drivers (> 4xxx) that use GLX 1.2. > GLX 1.2 is used by Mesa... Hmms that is strange, I'm on nvidia 100% positive... > > I'm implementing the VisInput ref counting right now, will check it > > into CVS. I ended up implementing VisObject and have refcounting and object dtoring for virtually every object in libvisual :) Cheers, Dennis |
|
From: Vitaly V. B. <vit...@us...> - 2004-11-24 17:29:42
|
On Tue, 23 Nov 2004 19:13:44 +0100 Dennis Smit <sy...@yo...> wrote: > Vitaly, I can't use the GLX drivers because it currently only supports > GLX 1.3, and the nvidia drivers I have are 1.2... What is the actual That's wrong. I know no nvidia drivers (> 4xxx) that use GLX 1.2. GLX 1.2 is used by Mesa... > difference between the two ? (in this case that is)... some interfaces are bit different. Context management/offscreen rendering for example. > I'm implementing the VisInput ref counting right now, will check it > into CVS. OK. -- Vitaly GPG Key ID: F95A23B9 |
|
From: Dennis S. <sy...@yo...> - 2004-11-24 00:54:44
|
Suggestions ?, we could have an extra entry in the object instance
structure..
Cheers,
Dennis
On Tue, 2004-11-23 at 20:43 -0300, Duilio Javier Protti wrote:
> The problem is not G_DEFINE_TYPE, which could be easily defined, but
> G_TYPE_INSTANCE_GET_PRIVATE, which needs
> g_type_class_{add/get}_private() functions. I think we need another way
> to save the private info.
>
>
> Bye,
> Duilio.
>
>
>
>
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://productguide.itmanagersjournal.com/
> _______________________________________________
> Libvisual-devel mailing list
> Lib...@li...
> https://lists.sourceforge.net/lists/listinfo/libvisual-devel
|
|
From: Duilio J. P. <dp...@fc...> - 2004-11-23 23:41:53
|
The problem is not G_DEFINE_TYPE, which could be easily defined, but
G_TYPE_INSTANCE_GET_PRIVATE, which needs
g_type_class_{add/get}_private() functions. I think we need another way
to save the private info.
Bye,
Duilio.
|
|
From: Dennis S. <sy...@yo...> - 2004-11-23 23:21:40
|
Alright, most of the libvisual system is now in the VisObject fashion. This introduces the following API functions: VisObject *visual_object_new (void); int visual_object_free (VisObject *object); int visual_object_destroy (VisObject *object); int visual_object_ref (VisObject *object); int visual_object_unref (VisObject *object); int visual_object_set_private (VisObject *object, void *priv); void *visual_object_get_private (VisObject *object); Nearly every libvisual structure (like VisActor / VisSongInfo etc etc) can be cast into a VisObject using object = VISUAL_OBJECT (actor); This way we provide a generalized ref counting interface. That also supports complex object desctructors that are set at object creation. With this change most of the other _free and _destroy functions are history and object disposing should be done using an visual_object_unref, and if you're really sure visual_object_destroy. To get this working, EVERY libvisual module needs a recompile. I haven't updated lvdisplay to this, Vitaly would you like to take a look at VisObjectifying lvdisplay ? VisUI and lv_plugin.c still needs to be done, I will most likely do that tomorrow in the evening. Cheers, Dennis |
|
From: Duilio J. P. <dp...@fc...> - 2004-11-23 23:07:36
|
I'm having big troubles here, because G_TYPE_INSTANCE_GET_PRIVATE and G_DEFINE_TYPE macros are not defined on gtk2 < 2.4, so I must provide one, but first I will take a look at the semanthic of this things... Bye, Duilio. > Would be EXCELENT!! :) > > Thanks! > > > On Tue, 2004-11-23 at 01:30 -0300, Duilio Javier Protti wrote: > > Ok, I think it's a good idea. I can send a patch to configure.ac and > > lvw_gtk2_visui.c to work around this. > > > > > > Bye, > > Duilio. |
|
From: Dennis S. <sy...@yo...> - 2004-11-23 18:14:05
|
Vitaly, I can't use the GLX drivers because it currently only supports GLX 1.3, and the nvidia drivers I have are 1.2... What is the actual difference between the two ? (in this case that is)... I'm implementing the VisInput ref counting right now, will check it into CVS. Cheers, Dennis |
|
From: Dennis S. <sy...@yo...> - 2004-11-23 10:41:54
|
Would be EXCELENT!! :) Thanks! On Tue, 2004-11-23 at 01:30 -0300, Duilio Javier Protti wrote: > Ok, I think it's a good idea. I can send a patch to configure.ac and > lvw_gtk2_visui.c to work around this. > > > Bye, > Duilio. |
|
From: Dennis S. <sy...@yo...> - 2004-11-23 10:41:36
|
I had already fixed that in CVS! :)
But thanks for noticing
On Tue, 2004-11-23 at 01:30 -0300, Duilio Javier Protti wrote:
> Maybe there is an error on visual_thread_join() function, on lines:
>
> if (pthread_join (&thread->thread, &result) < 0) {
> visual_log (VISUAL_LOG_CRITICAL, "Error while joining thread");
> ...
>
> I think we must not take the address of thread->thread, because
> pthread_join()'s prototype is:
>
> int pthread_join (pthread_t __th, void **__thread_return);
>
> where we see that the first argument is not a pointer to a pthread_t.
>
> But maybe I'm wrong...
>
>
> Bye,
> Duilio.
>
>
>
>
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://productguide.itmanagersjournal.com/
> _______________________________________________
> Libvisual-devel mailing list
> Lib...@li...
> https://lists.sourceforge.net/lists/listinfo/libvisual-devel
|
|
From: Duilio J. P. <dp...@fc...> - 2004-11-23 04:28:28
|
Ok, I think it's a good idea. I can send a patch to configure.ac and lvw_gtk2_visui.c to work around this. Bye, Duilio. > I think it's best to use GtkOptionMenu with gtk < 2.4 > > and I bet there is something like that for gtk 1.2 as well :) > > What do you think ? |
|
From: Duilio J. P. <dp...@fc...> - 2004-11-23 04:28:28
|
Maybe there is an error on visual_thread_join() function, on lines:
if (pthread_join (&thread->thread, &result) < 0) {
visual_log (VISUAL_LOG_CRITICAL, "Error while joining thread");
...
I think we must not take the address of thread->thread, because
pthread_join()'s prototype is:
int pthread_join (pthread_t __th, void **__thread_return);
where we see that the first argument is not a pointer to a pthread_t.
But maybe I'm wrong...
Bye,
Duilio.
|
|
From: Dennis S. <sy...@yo...> - 2004-11-22 22:45:43
|
Some random thoughts on VisObject:
Object hierachy:
VisObject
VisUI
VisUI...
VisActor
VisInput
VisPlugin
etc
Implement Ref counting at VisObject layer.
Most important API addition for now:
visual_object_ref (VisObject *object);
visual_object_unref (VisObject *object);
VisObject {
int refcount;
VisObjectDtorFunc dtor;
}
Example for a random structure:
VisActor {
VisObject object;
blah;
blah;
}
visual_actor_new ()
{
VisActor *actor = ...
VISUAL_OBJECT (actor)->dtor = dtor_actor;
}
visual_object_unref (obj)
{
obj->ref--
if (obj->ref <= 0)
obj->dtor (obj);
}
Problems to be solved:
How to implement the automatic dtorring.
Being able to set a dtor for an object ?
This could also exterminate the ugly switches in VisUI that are used for
a VisUI object dtoring.
Please comment on this :)
|
|
From: Dennis S. <sy...@yo...> - 2004-11-22 19:59:46
|
We don't use numeric plugin type indicators any longer. This means that we did a major API breakage and when CVS is checked out EVERY libvisual module needs to be recompiled (make clean ; make install) What does this mean from a technical point of view. The numeric plugin types gave a problem because it was easy to collide between different libraries that register different plugin types. The new fashion is: "Domain:package:type", for actor plugins this means "Libvisual:core:actor" For lv-display frontend plugins this would mean "Libvisual:display:frontend", this scheme is very flexible and it's very easy to register new plugin types to the plugin loading system. We introduced a heap of helper functions regarding this, check it out at lv_plugin.h :) Cheers, Dennis |
|
From: Dennis S. <sy...@yo...> - 2004-11-22 17:58:44
|
Why would that be needed ?, libvisual should detect of GL is present or not. On Mon, 2004-11-22 at 19:01 +0000, salsaman wrote: > Hi, > is it possible to add a: pkg-config --libs-nogl to package config for > libvisual ? > This would make compiling libvis much easier. > > Thanks, > Salsaman. |
|
From: salsaman <sal...@xs...> - 2004-11-22 17:45:11
|
Hi, is it possible to add a: pkg-config --libs-nogl to package config for libvisual ? This would make compiling libvis much easier. Thanks, Salsaman. |
|
From: Dennis S. <sy...@yo...> - 2004-11-22 16:10:10
|
On Sun, 2004-11-21 at 22:38 +0200, Vitaly V. Bursov wrote:
> Hi! :)
HI!! where have you been!, I'm so glad to see that you're back !!!
*YAY!*
> lvdisplay now supports multiple screens :)
Neat! :)
> ======= example/main.c
> VisInput *input = visual_input_new("alsa");
> [snip]
> visual_bin_connect(bin, actor, input);
> visual_bin_connect(bin2, actor2, input);
> [snip]
> lvdisplay_finalize(v);
> lvdisplay_finalize(v2);
> ========
> yup, "input" gets free()'d 2 times. How we're going to fix it?
Ai, I see the problem!
> a. forbid such kind of code. Add input->is_used field to debug it.
> b. add refcounting to input and alike structures and free it accordingly
>
> I think b. is nice here. Because we have a reentrant design, plugin
> can be used multiple times.... or not?
I agree, we should introduce refcounting. I've got to add something to
this. What about introducing a VisObject structure, from which every
structure can 'inheritent' by making sure it's allocated in the top of
the struct like I did in VisUI and do ref counting things on a global
scale, something like visual_object_ref, visual_object_unref etc...
We could introduce cast macros so you get something like
VisInput *input = blahblah;
visual_object_unref (VISUAL_OBJECT (input)); to decrease the ref count.
What do you think ?, and are there things I should keep in mind, I
never implemented ref counting before, but I understand the idea...
> and BTW, why not to move -O3 flag to configure scripts and use it only
> when --enable-debug is not passed?
Please go ahead and do this, if you'd like :)
Cheers,
Dennis
|
|
From: Dennis S. <sy...@yo...> - 2004-11-22 15:53:33
|
I think it's best to use GtkOptionMenu with gtk < 2.4 and I bet there is something like that for gtk 1.2 as well :) What do you think ? On Sat, 2004-11-20 at 02:44 -0300, Duilio Javier Protti wrote: > >Do you have a suggestion for the work around combobox thingy ? > > > >Cheers, > >Dennis > > I see three possibilities here: > > - Avoid the use of GtkComboBox within libvisual-widgets. > - Use GtkComboBox, and give our own implementation when gtk2 < 2.4. > - Require gtk2 >= 2.4. > > The second option is not so much weird as it appears on a first look: if > we decide to use GtkComboBox, even if we require gtk2 >= 2.4, we will > need to implement combobox for the gtk1 port anyway, because combobox > doesn't exists on gtk1 and it will never do. > > IMHO, the first option is the best, but just a suggestion... > > > Bye, > Duilio. |
|
From: Dennis S. <sy...@yo...> - 2004-11-22 15:51:26
|
On Fri, 2004-11-19 at 19:55 +0000, salsaman wrote: > Hi, > some files for goom2 appears to be missing from 0.1.7. It does not get > installed on a fresh build. > > Cheers, > Salsaman. Install libgoom2 http://prdownloads.sourceforge.net/goom/goom2k4-dev21.tar.gz?download and reconfigure libvisual-plugins. Cheers. |
|
From: Vitaly V. B. <vit...@us...> - 2004-11-21 20:39:47
|
Hi! :)
lvdisplay now supports multiple screens :)
...but before you start demo program...
there's a problem.
=======
*** glibc detected *** double free or corruption: 0x0810a338 ***
[ that's my new glibc :) I like it! ]
Program received signal SIGABRT, Aborted.
[Switching to Thread 1083998352 (LWP 4063)]
0xffffe410 in ?? ()
(gdb) bt
[ snip ]
#8 0x40029df1 in visual_mem_free (ptr=0x0) at lv_mem.c:49
#9 0x40026c41 in visual_plugin_unload (plugin=0x1fa) at lv_plugin.c:504
#10 0x40022e1e in visual_input_destroy (input=0x805fd40) at lv_input.c:163
#11 0x40023c08 in visual_bin_destroy (bin=0x805fcc8) at lv_bin.c:91
#12 0x40018bf5 in lvdisplay_finalize (v=0x8057300) at lv_display.c:326
#13 0x08048de5 in main (argc=1, argv=0xbffff6a4) at main.c:243
=======
======= example/main.c
VisInput *input = visual_input_new("alsa");
[snip]
visual_bin_connect(bin, actor, input);
visual_bin_connect(bin2, actor2, input);
[snip]
lvdisplay_finalize(v);
lvdisplay_finalize(v2);
========
yup, "input" gets free()'d 2 times. How we're going to fix it?
a. forbid such kind of code. Add input->is_used field to debug it.
b. add refcounting to input and alike structures and free it accordingly
I think b. is nice here. Because we have a reentrant design, plugin
can be used multiple times.... or not?
and BTW, why not to move -O3 flag to configure scripts and use it only
when --enable-debug is not passed?
--
Vitaly
GPG Key ID: F95A23B9
|
|
From: Duilio J. P. <dp...@fc...> - 2004-11-20 05:43:28
|
>Do you have a suggestion for the work around combobox thingy ? > >Cheers, >Dennis I see three possibilities here: - Avoid the use of GtkComboBox within libvisual-widgets. - Use GtkComboBox, and give our own implementation when gtk2 < 2.4. - Require gtk2 >= 2.4. The second option is not so much weird as it appears on a first look: if we decide to use GtkComboBox, even if we require gtk2 >= 2.4, we will need to implement combobox for the gtk1 port anyway, because combobox doesn't exists on gtk1 and it will never do. IMHO, the first option is the best, but just a suggestion... Bye, Duilio. |
|
From: salsaman <sal...@xs...> - 2004-11-19 18:39:19
|
Hi, some files for goom2 appears to be missing from 0.1.7. It does not get installed on a fresh build. Cheers, Salsaman. |
|
From: Dennis S. <sy...@yo...> - 2004-11-19 07:14:41
|
Alright, thank you !! :) Do you have a suggestion for the work around combobox thingy ? Cheers, Dennis On Fri, 2004-11-19 at 00:20 -0300, Duilio Javier Protti wrote: > I have changed just a bit the configure.ac, basicly removing some unused > checks. > > I could not test the package yet because it doesn't build on my system, > because the use of GtkComboBox, which was introduced on gtk+-2.3 (I'm > using RedHat 9 with gtk+ 2.2.2). I think we must work around this, > because there is so much code out there with gtk+ < 2.4 things. > > > Bye, > Duilio. > > > > > ------------------------------------------------------- > This SF.Net email is sponsored by: InterSystems CACHE > FREE OODBMS DOWNLOAD - A multidimensional database that combines > robust object and relational technologies, making it a perfect match > for Java, C++,COM, XML, ODBC and JDBC. www.intersystems.com/match8 > _______________________________________________ > Libvisual-devel mailing list > Lib...@li... > https://lists.sourceforge.net/lists/listinfo/libvisual-devel |
|
From: Dennis S. <sy...@yo...> - 2004-11-19 06:16:16
|
Thanks a lot, I'll check it into CVS, great :) On Fri, 2004-11-19 at 01:32 -0300, Duilio Javier Protti wrote: > Ok, here is a patch for a basic detection of thread implementation. > Actually, linker flags are just set to -lpthread, but in future I will > detect that too. > > > Bye, > Duilio. > |
|
From: Duilio J. P. <dp...@fc...> - 2004-11-19 04:31:03
|
Ok, here is a patch for a basic detection of thread implementation. Actually, linker flags are just set to -lpthread, but in future I will detect that too. Bye, Duilio. |
10 messages has been excluded from this view by a project administrator.
