Nabble - k3d-development

Re: Delete components

2008-10-12T07:28:00Z

Bart Janssens wrote:

> I'm looking at converting the Delete plugin to the new mesh structure.
> In the old version, a face got deleted when an edge or its companion
> belonging to that face was selected. In the current state of affairs,
> the UI makes sure that when you select an edge, both it and its
> companion are selected. This is not necessarily the case with loaded
> documents.
>
> It does not seem right to keep the old behavior of Delete, where the
> selected companion triggers deletion of the adjacent face. This would
> result in unexpected behavior when we allow explicit selection of
> half-edges (one would expect only one face to get deleted, not both
> faces on either side of the half-edge). One way I see to retain
> compatibility, is to convert the mesh selection when a Delete plugin
> gets loaded, so both edges and their companions are selected. Is this
> OK?

You might want to confirm with Joe that the current Delete behavior is
correct - I'm pretty sure it follows his requirements, but this is an
opportunity to double-check and alter any behaviors that are suboptimal.

As always, plugins should be dumb and do exactly what they're told, and
your proposal is consistent with that. The only thing you haven't
mentioned is that at load-time you will need to differentiate between
"old" selections that should be "upgraded" and "new" selections that
shouldn't. My suggestion is that you create a new "Delete" plugin with
new uuid. That means your load logic will be:

if(Old Delete node)
{
Convert old selection.
Convert old uuid to new uuid.
}
else
{
Do nothing.
}

Cheers,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Delete components

2008-10-12T02:14:02Z

Hi gang,

I'm looking at converting the Delete plugin to the new mesh structure.
In the old version, a face got deleted when an edge or its companion
belonging to that face was selected. In the current state of affairs,
the UI makes sure that when you select an edge, both it and its
companion are selected. This is not necessarily the case with loaded
documents.

It does not seem right to keep the old behavior of Delete, where the
selected companion triggers deletion of the adjacent face. This would
result in unexpected behavior when we allow explicit selection of
half-edges (one would expect only one face to get deleted, not both
faces on either side of the half-edge). One way I see to retain
compatibility, is to convert the mesh selection when a Delete plugin
gets loaded, so both edges and their companions are selected. Is this
OK?

Cheers,

--
Bart

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Missing Control Vertices

2008-10-04T15:26:32Z

Esben Stien wrote:
> "Carsten Haubold" <CarstenHaubold@...> writes:
>
>> you can find the code for drawing in modules/opengl_painters/
>> point_painter.cpp
>
> Yes, I see the tiny infinitesimal little pixel. If I move the object
> extremely slowly it will appear before seemingly being hidden again. I
> got a 1920x1200 resolution on my monitor.
AFAIK you can change the size from the "GL Engine" node.
The Point Size property.

Cheers!
Joaquín

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Plans for the Summit?

2008-10-01T14:18:56Z

On Tue, Sep 30, 2008 at 5:10 PM, Joaquin Duo <hoakoduo@...> wrote:
> Is there any topic we should try to bring to the summit?
> Any plan, suggestion, comment?

On Sunday, there are "hackfest" sessions. I assume this means we will
be working on K-3D, perhaps this requires some coordination to see
what we do? One idea is to turn it into a bug day.

> I send this to the mailing list so its open to the community suggestions.
> Specially the Gsoc students could comment on this.

I don't have any concrete ideas at the moment for the GSoC "lessons
learned" sessions, given that (from my perspective, anyway) everything
went great this year. I agree that it would be nice to hear what the
students think could be improved in the future. Note that everything
including the selection process is up for discussion, so anyone who
did apply but was not selected, feel free to chime in!

One final note: it seems to be normal that the schedule is still
vague, last year the agenda was decided upon the day itself. It does
help to think about it in advance, though :)

Cheers,

--
Bart

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: attribute_array_copier::mix()

2008-09-30T22:16:15Z

Bart Janssens wrote:

> Either way, I noticed the current attribute array implementation in
> CatmullClark is still wrong, and the mix method is completely
> unnecessary: I can just use an attribute array copier that has the
> same input and output array for this specific use case.

Yep, that's a much better solution.

Cheers,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Missing Control Vertices

2008-09-30T18:06:21Z

"Carsten Haubold" <CarstenHaubold@...> writes:

> you can find the code for drawing in modules/opengl_painters/
> point_painter.cpp

Yes, I see the tiny infinitesimal little pixel. If I move the object
extremely slowly it will appear before seemingly being hidden again. I
got a 1920x1200 resolution on my monitor.

This is really frustrating. Can we have an option to draw a bigger
point?. I would really like to have points the size of that little
sphere in the center of the moving tool. That's a nice looking point.

--
Esben Stien is b0ef@e s a
http://www. s t n m
irc://irc. b - i . e/%23contact
sip:b0ef@ e e
jid:b0ef@ n n

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: attribute_array_copier::mix()

2008-09-30T13:42:11Z

On Tue, Sep 30, 2008 at 3:05 AM, Timothy M. Shead <tshead@...> wrote:
> Just noticed this addition, which is going to have to be removed -
> because it modifies the same arrays that it uses as inputs, the order in
> which you visit the target values affects the outcome, and it's
> virtually impossible to guarantee usable results. What's the
> performance issue with copy()?

The issue is specific for CatmullClark: edge vertex and corner vertex
positions both depend on face vertex positions, and the same happens
when calculating the other attribute data. It is therefore necessary
to calculate a weighted sum involving previously calculated face data,
which is already in the output array. Previously, I simple
recalculated the data when needed again, but this almost doubled the
running time of the entire CatmullClark operation.

Either way, I noticed the current attribute array implementation in
CatmullClark is still wrong, and the mix method is completely
unnecessary: I can just use an attribute array copier that has the
same input and output array for this specific use case.

Correctness is (well, will be :) guaranteed because all face data is
calculated in a first sweep, then edge data in a second sweep, and
finally point data in a last sweep. This is necessary anyway for
future parallelization.

Cheers,

--
Bart

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Plans for the Summit?

2008-09-30T08:10:23Z

Hi folks!

Is there any topic we should try to bring to the summit?
Any plan, suggestion, comment?

I send this to the mailing list so its open to the community suggestions.
Specially the Gsoc students could comment on this.

Cheers!
Joaquín

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

attribute_array_copier::mix()

2008-09-29T18:05:43Z

Bart:

Just noticed this addition, which is going to have to be removed -
because it modifies the same arrays that it uses as inputs, the order in
which you visit the target values affects the outcome, and it's
virtually impossible to guarantee usable results. What's the
performance issue with copy()?

Cheers,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Pipeline_data and MergeMesh

2008-09-29T14:19:26Z

On Mon, Sep 29, 2008 at 12:52 AM, Timothy M. Shead <tshead@...> wrote:
> No, because I explicitly wish to discourage the style of coding where
> the developer doesn't know whether they created something or not - it's
> sloppy, difficult to troubleshoot, and never actually necessary.

OK, I agree!

> current implementation completely obsolete. I think the expedient thing
> is to make create_if_not_exists() a non-member template function that is
> a private implementation detail of MergeMesh - it will be an easy
> search-and-replace to put it to use.

Yep, it was relatively painless ;)

Cheers,

--
Bart

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Document Loading Times

2008-09-28T15:55:52Z

Joaquin Duo wrote:
> Timothy M. Shead wrote:

> Also seems is not the amount of plugins only, but the type, since i
> tried 200 BulgePoints modifiers and has no effect.

>> See
>> modules/test/pipeline_profiler.cpp for some sample code that could be
>> temporarily copied to modules/k3d_io/k3d_document_importer.cpp for testing.

> After all, should I made this test to see if the process is eating CPU
> inside the node evaluation? Or the information above gives enough
> detail? (ill be trying to do it anyway)

Synthesizing change events shouldn't cause any execution at all, that's
why I'd like to see the results from the pipeline profiler - if some
node is executing when it receives a change event, that's a bug.

Cheers,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Pipeline_data and MergeMesh

2008-09-28T15:52:31Z

Bart Janssens wrote:

> I was wondering if we could add the following method to pipeline_data.h:
>
> T& create_if_not_exists()
> {
> if(!storage)
> {
> return create();
> }
> else
> {
> return writable();
> }
> }

No, because I explicitly wish to discourage the style of coding where
the developer doesn't know whether they created something or not - it's
sloppy, difficult to troubleshoot, and never actually necessary.

> It would allow simply replacing all the calls to create() in
> MergeMesh, which does not work at the moment, because the already
> merged data gets discarded.

I can think of two rearrangements of the outer-loop in MergeMesh that
would eliminate the need for this logic altogether, but I'm not
advocating this approach since generic meshes will eventually make any
current implementation completely obsolete. I think the expedient thing
is to make create_if_not_exists() a non-member template function that is
a private implementation detail of MergeMesh - it will be an easy
search-and-replace to put it to use.

Cheers,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Document Loading Times

2008-09-28T15:04:13Z

Timothy M. Shead wrote:

>> Obviously the problem was on the connections between properties since removing
>> them from a K-3D document change the times on an unpatched K-3D.
>
> No question that this is a problem that needs to be solved, but this may
> be a sign of a deeper issue. Simply setting property dependencies
> should not take this much time - either storing undo/redo information
> when properties change is out-of-control (doubtful), or the pipeline is
> being executed, which in-theory should never happen. It would be useful
> if you could capture some information on whether this is the case - you
> can do a little hacking on the serialization code so that it listens to
> the pipeline profiler interface and prints execution times to the log -
> again, there shouldn't be any, but probably is.

I made first a simple test and this limes seem to be the problem.

// Synthesize change notifications for every property whose parent was
set ...
for(dependencies_t::iterator dependency = Dependencies.begin();
dependency != Dependencies.end(); ++dependency)
dependency->first->property_changed_signal().emit(Hint);

Even if i change the last line to
dependency->first->property_changed_signal().emit(0);
There is a lot of delay on loading.

Also seems is not the amount of plugins only, but the type, since i
tried 200 BulgePoints modifiers and has no effect.

Maybe are the plugins that recieve hints?

> See
> modules/test/pipeline_profiler.cpp for some sample code that could be
> temporarily copied to modules/k3d_io/k3d_document_importer.cpp for testing.
After all, should I made this test to see if the process is eating CPU
inside the node evaluation? Or the information above gives enough
detail? (ill be trying to do it anyway)

Cheers!
Joaquín

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Pipeline_data and MergeMesh

2008-09-28T13:26:52Z

Tim:

I was wondering if we could add the following method to pipeline_data.h:

T& create_if_not_exists()
{
if(!storage)
{
return create();
}
else
{
return writable();
}
}

It would allow simply replacing all the calls to create() in
MergeMesh, which does not work at the moment, because the already
merged data gets discarded.

Cheers,

--
Bart

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: SubdivideEdges

2008-09-28T03:27:44Z

On Sat, Sep 27, 2008 at 7:10 PM, Timothy M. Shead <tshead@...> wrote:
> Makes sense - I suspect that the same thing is happening in
> SubdivideFaces, could you take a look?

OK, got it working again :) I'll probably be revisiting SubdivideEdges
and -Faces some more after the new CatmullClark filter is complete, to
improve consistency and readability in the code.

Cheers,

--
Bart

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Document Loading Times

2008-09-28T01:58:15Z

Timothy M. Shead wrote:

Re: SubdivideEdges

2008-09-27T10:05:31Z

Bart Janssens wrote:

> On Sat, Sep 27, 2008 at 2:53 AM, Timothy M. Shead <tshead@...> wrote:
>> Bart:
>>
>> Could you take a look at SubdivideEdges? I'm drawing a blank on why
>> some of the tests are failing. See
>
> OK, fixed in SVN! The problem was that I was using some arrays from
> output_polyhedra, that using the old shared_ptr system still had data
> from the source mesh, but with the new pipeline_data system already
> contained new (zero) data.

Makes sense - I suspect that the same thing is happening in
SubdivideFaces, could you take a look?

Cheers,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: SubdivideEdges

2008-09-27T05:15:08Z

On Sat, Sep 27, 2008 at 2:53 AM, Timothy M. Shead <tshead@...> wrote:
> Bart:
>
> Could you take a look at SubdivideEdges? I'm drawing a blank on why
> some of the tests are failing. See

OK, fixed in SVN! The problem was that I was using some arrays from
output_polyhedra, that using the old shared_ptr system still had data
from the source mesh, but with the new pipeline_data system already
contained new (zero) data.

Cheers,

--
Bart

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

SubdivideEdges

2008-09-26T17:49:26Z

Bart:

Could you take a look at SubdivideEdges? I'm drawing a blank on why
some of the tests are failing. See

http://k3d.svn.sourceforge.net/viewvc/k3d/trunk/modules/mesh/subdivide_edges.cpp?r1=1381&r2=1459

to see what's changed.

Many thanks,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Quiet out there

2008-09-26T13:59:31Z

Alex Curtis wrote:
> Hi gang,
>
> Sorry for my absence. I have been so busy moving into a new flat /
> starting University again.
>
> I should be more vocal / useful from now on.

Good to hear from you there, Amigo! You should brief people on your
ideas for a render engine, I think everyone would be interested ...

Cheers,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Quiet out there

2008-09-24T13:23:36Z

Hi gang,

Sorry for my absence. I have been so busy moving into a new flat /
starting University again.

I should be more vocal / useful from now on.

Cheers
-Alex

Timothy M. Shead wrote:

> Bart Janssens wrote:
>
>> On 9/7/08, Timothy M. Shead <tshead@...> wrote:
>>
>>
>>> * I've been making good progress on generic mesh primitives:
>>>
>>> http://www.k-3d.org/wiki/Generic_Mesh_Primitives#Implementation
>>>
>>>
>>>
>> Looks very promising! I assume that once this is complete, the current
>> array meshes will have to co-exist with the generic mesh primitives
>> until all other plugins have been converted? At least it should be a
>> whole lot more straightforward than the (on-going) legacy::mesh to
>> array-mesh conversion :
>>
> That's the idea.
>
> Cheers,
> Tim
>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> _______________________________________________
> K3d-development mailing list
> K3d-development@...
> https://lists.sourceforge.net/lists/listinfo/k3d-development
>
>

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: K-3D crash on document loading

2008-09-21T20:10:21Z

Joaquin Duo wrote:
> Hi again!
>
> Seems that my 3D object is quite picky. :-D
>
> http://sourceforge.net/tracker/index.php?func=detail&aid=2101862&group_id=11113&atid=311113
>
> After updating from SVN i got the following crash while loading it

Check the dashboard, there are still many regressions to be cleaned-up.

Cheers,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

How about linux?

2008-09-21T01:31:51Z

Hello dear creators and developers of k3d, finally i found you!
In the k3d wiki page I read "GNU/Linux is the main development platform and preferred operating environment for K-3D"
And there's always newest windows and mac build, and linux have frozen at .6.7 version.
How can you explain that? Maibe just update the page, and say that linux users can compile it, but not very welcome?
And more, I tried to compile it on debian 64bit, followed your instructions, and while I run it with make run it runs, but when I installing it (in /usr/local) and then run , it fails to find libraries. Please, update instructions how to configure everything. It's not enough to give a list of distributions, and then send them all to cmake build, where you say "Follow the on-screen instructions to configure the build"
I understand, that it's not easy to develop program, and same time write instructions, at least it's boring... Then why don't you start users forum, and give a link to it on main page? there's no information about developers, or where i can give my stupid questions, and have a stupid answer to it! look at Blenders forums, such uncomfortable program, and such success! Because users help each other, and even help to develop blender,because developers know users needs. look at powray, same situation. Users talk each other, sharing files, THEY ARE A LIVE!!! And your forum is hidden, I could find it only by googling... Where K3D users live? Where's examples of successful work, tutorials (not only how to build a snow ball)
Or k3d is your (developers) toy, not oriented at the end user? You are not interested to use it in "real life"? Then you will die, like many many wonderful projects before you died!
P.S. Sorry, I'm a bit angry, and my English bad...

K-3D crash on document loading

2008-09-20T04:08:19Z

Hi again!

Seems that my 3D object is quite picky. :-D

http://sourceforge.net/tracker/index.php?func=detail&aid=2101862&group_id=11113&atid=311113

After updating from SVN i got the following crash while loading it

Cheers!
Joaquín

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x40ff0ce0 (LWP 25571)]
0x4051d2cb in k3d::operator- () from ...k3d_build/lib/libk3dsdk.so
(gdb) t
[Current thread is 1 (Thread 0x40ff0ce0 (LWP 25571))]
(gdb) bt
#0 0x4051d2cb in k3d::operator- () from ...k3d_build/lib/libk3dsdk.so
#1 0x4535c301 in
module::mesh::detail::split_point_calculator::operator() ()
from ...k3d_build/lib/k3d/plugins/k3d-mesh.module
#2 0x4535c7db in module::mesh::subdivide_edges::on_update_mesh ()
from ...k3d_build/lib/k3d/plugins/k3d-mesh.module
#3 0x405be6a9 in k3d::mesh_modifier<k3d::node>::initialize_mesh () from
...k3d_build/lib/libk3dsdk.so
#4 0x405b81e6 in sigc::bound_mem_functor1<void,
k3d::mesh_modifier<k3d::node>, k3d::mesh&>::operator() ()
from ...k3d_build/lib/libk3dsdk.so
#5 0x405b8213 in sigc::adaptor_functor<sigc::bound_mem_functor1<void,
k3d::mesh_modifier<k3d::node>, k3d::mesh&> >::operator()<k3d::mesh&> ()
from ...k3d_build/lib/libk3dsdk.so
#6 0x405b8247 in
sigc::internal::slot_call1<sigc::bound_mem_functor1<void,
k3d::mesh_modifier<k3d::node>, k3d::mesh&>, void, k3d::mesh&>::call_it
() from ...k3d_build/lib/libk3dsdk.so
#7 0x405b8a6f in sigc::slot1<void, k3d::mesh&>::operator() () from
...k3d_build/lib/libk3dsdk.so
#8 0x405bdb9f in k3d::data::pointer_storage<k3d::mesh*,
k3d::data::change_signal<k3d::mesh*> >::internal_value ()
from ...k3d_build/lib/libk3dsdk.so
#9 0x405bdd31 in k3d::data::read_only_property<k3d::mesh*,
k3d::data::immutable_name<k3d::data::no_constraint<k3d::mesh*,
k3d::data::no_undo<k3d::mesh*, k3d::data::pointer_storage<k3d::mesh*,
k3d::data::change_signal<k3d::mesh*> > > > > >::property_internal_value

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Sharing the Pain ...

2008-09-17T18:00:18Z

Hi again!

> * Carlos/Joaquin: Similarly, I'm having getting the
> Collada dependency
> setup.
Ill check this after Friday, I have another exam that day.

Cheers!
Joaquín

Yahoo! Cocina
Recetas prácticas y comida saludable
http://ar.mujer.yahoo.com/cocina/

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Sharing the Pain ...

2008-09-16T22:36:13Z

Bart Janssens wrote:
> On 9/16/08, Timothy M. Shead <tshead@...> wrote:
>> See some new documentation that I put together on these changes:
>>
>> http://www.k-3d.org/wiki/Pipeline_Data
>> http://www.k-3d.org/wiki/Named_Array_Collections
>
> Cool, this seems much easier to use! In the end of the Pipeline Data
> page, there seems to be a missing code sample after "replace code of
> the form". I think a call to k3d::make_unique should be there.

Good catch, I've fixed this and both documents are now more complete.

> This reminds me: shouldn't we move NewTweakPoints out of the
> development module and have it replace the old (unused, normally)
> TweakPoints?

Yes, the original TweakPoints should be deprecated and renamed to
LegacyTweakPoints, so the new version can take its place.

Cheers,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Sharing the Pain ...

2008-09-16T14:43:50Z

On 9/16/08, Timothy M. Shead <tshead@...> wrote:
> See some new documentation that I put together on these changes:
>
> http://www.k-3d.org/wiki/Pipeline_Data
> http://www.k-3d.org/wiki/Named_Array_Collections

Cool, this seems much easier to use! In the end of the Pipeline Data
page, there seems to be a missing code sample after "replace code of
the form". I think a call to k3d::make_unique should be there.

> * Bart: I haven't gotten OpenCascade setup on my machine, can you look
> into the OpenCascade module?

Yes. I'll postpone this one until I get round to some (half-finished)
further tweaking of the OpenCascade converter, as some primitives
still aren't processed correctly.

> * Bart: I'm not clear on what MakeSharpEdges is supposed to be doing -
> it modifies its' own property in on_create_mesh(), then uses the
> property value in on_update_mesh(), what's that about? It seems
> experimental, can we get rid of it?

This was an experiment to get a plugin to control face normals based
on sharp and non-sharp edges. I have removed it for now, as I never
really got anywhere with it. I think the basic idea was to keep a
sharp-edge list, and use that to update a named normal array. In
retrospect, this edge list is not needed, and the selection array
should be used instead.

> * Bart: I don't trust myself to modify NewTweakPoints - something I hope
> to fix with the completion of the hint-mapping changes, but in the
> meantime ...

Fixed in SVN!

This reminds me: shouldn't we move NewTweakPoints out of the
development module and have it replace the old (unused, normally)
TweakPoints?

You may have noticed the lack of dashboard submissions from my
machine. The problem is that my own internet connection is down,
hopefully it will be fixed tomorrow (until then, there's the laptop
and an unknow(n)(ing) kind neighbour :)

Cheers,

--
Bart

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Sharing the Pain ...

2008-09-15T23:00:19Z

So, the work to convert from boost::shared_ptr to k3d::pipeline_data is
going-well, and I am happy with the way it eliminates a lot of redundant
code and provides better clarity on what's happening.

See some new documentation that I put together on these changes:

http://www.k-3d.org/wiki/Pipeline_Data
http://www.k-3d.org/wiki/Named_Array_Collections

In particular, we have been using k3d::make_unique() for a couple of
unrelated operations - as a way to get a writable reference to an
existing array and as a way to create new arrays, which is pretty
misleading. The changes eliminate this in favor of some idioms that are
both more explicit and involve fewer lines of code.

The down-side is that this has grown into a massive project that touches
nearly everything, which (fortunately) happens pretty rarely these days.
Normally, I wouldn't commit these changes until every test was passing,
but this time around, I really can't get it all done alone. In particular:

* Bart: I haven't gotten OpenCascade setup on my machine, can you look
into the OpenCascade module?

* Carlos/Joaquin: Similarly, I'm having getting the Collada dependency
setup.

* Bart: I'm not clear on what MakeSharpEdges is supposed to be doing -
it modifies its' own property in on_create_mesh(), then uses the
property value in on_update_mesh(), what's that about? It seems
experimental, can we get rid of it?

* Bart: I don't trust myself to modify NewTweakPoints - something I hope
to fix with the completion of the hint-mapping changes, but in the
meantime ...

I will be working on the remaining test failures, which shouldn't take
too long. The main gotcha at this point is calls to writable() that
should have been calls to create(), which is easy to diagnose and fix.

Many thanks,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Viewports

2008-09-13T15:25:13Z

> Anders Stenberg wrote:
>
>> I continue my trend of burst-development, it seems, and I'm still in a
>> off-phase. :)
>> However, I wrote up something about the viewport:
>> http://www.k-3d.org/wiki/Viewport_Refactoring
>>
>> Even though I've implemented large parts of it already, it's by no
>> means finished or bolted down. Feel free to give all kinds of input.
>> Code can always be changed. :)
>
> Sorry for the slow response on this, I think it's a great start, I might
> carve some of this functionality into additional interfaces, but I'm
> still restoring my system (decided to try Ubuntu for variety, so it's a
> learning process, too). I'll provide real feedback soon ...
>

I'm a bit stuck in other projects for the moment so I haven't been
able to work on it myself either. But I don't think it's that much
work to get the remaining parts (transform targets) working. I'll see
if I can try that soon.

/Anders

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Viewports

2008-09-13T15:18:56Z

Anders Stenberg wrote:

> I continue my trend of burst-development, it seems, and I'm still in a
> off-phase. :)
> However, I wrote up something about the viewport:
> http://www.k-3d.org/wiki/Viewport_Refactoring
>
> Even though I've implemented large parts of it already, it's by no
> means finished or bolted down. Feel free to give all kinds of input.
> Code can always be changed. :)

Sorry for the slow response on this, I think it's a great start, I might
carve some of this functionality into additional interfaces, but I'm
still restoring my system (decided to try Ubuntu for variety, so it's a
learning process, too). I'll provide real feedback soon ...

Cheers,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Document Loading Times

2008-09-13T15:16:43Z

Joaquín Duo wrote:

>> I'm rebuilding my workstation after some hardware problems, so I haven't
>> seen the patch, but this seems reasonable. What is the difference in
>> load-time that you are seeing?
> From my tests seems that loading time rises quadratically to amount of
> modifiers in an object's pipeline.
> To the extreme of taking more than 20 minutes for a document to load and having
> to kill the K-3D process. **
> After the patch it was almost instantaneous. (about 3 seconds)
>
> (** but i haven't sent that model, instead a model that takes about a few
> minutes without the patch)
>
> Obviously the problem was on the connections between properties since removing
> them from a K-3D document change the times on an unpatched K-3D.

No question that this is a problem that needs to be solved, but this may
be a sign of a deeper issue. Simply setting property dependencies
should not take this much time - either storing undo/redo information
when properties change is out-of-control (doubtful), or the pipeline is
being executed, which in-theory should never happen. It would be useful
if you could capture some information on whether this is the case - you
can do a little hacking on the serialization code so that it listens to
the pipeline profiler interface and prints execution times to the log -
again, there shouldn't be any, but probably is. See
modules/test/pipeline_profiler.cpp for some sample code that could be
temporarily copied to modules/k3d_io/k3d_document_importer.cpp for testing.

Cheers,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Document Loading Times

2008-09-12T02:33:21Z

Hi!

Timothy M. Shead wrote:
> I'm rebuilding my workstation after some hardware problems, so I haven't
> seen the patch, but this seems reasonable. What is the difference in
> load-time that you are seeing?
From my tests seems that loading time rises quadratically to amount of
modifiers in an object's pipeline.
To the extreme of taking more than 20 minutes for a document to load and having
to kill the K-3D process. **
After the patch it was almost instantaneous. (about 3 seconds)

(** but i haven't sent that model, instead a model that takes about a few
minutes without the patch)

Obviously the problem was on the connections between properties since removing
them from a K-3D document change the times on an unpatched K-3D.

Cheers!
Joaquín

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Document Loading Times

2008-09-11T21:23:43Z

Joaquín Duo wrote:

> I submitted a patch for speeding up document load. Specially for objects with a
> lot of modifiers.
>
> http://sourceforge.net/tracker/index.php?func=detail&aid=2101862&group_id=11113&atid=311113
>
> I haven't dig too deep
> On k3dsdk/serialization_xml.cpp file line 1401
> Document.pipeline().set_dependencies(dependencies);
> Seems to evaluate properties on every dependency update.
> This slows down considerably on document loading
> Thus i send a patch for this case with a method for this special case.
> Document.pipeline().set_dependencies_load_document(dependencies);
>
> Copying and commenting out the following lines in a new
> set_dependencies_load_document() method seems to solve the problem.

I'm rebuilding my workstation after some hardware problems, so I haven't
seen the patch, but this seems reasonable. What is the difference in
load-time that you are seeing?

Cheers,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Document Loading Times

2008-09-11T09:00:17Z

Hi Folks!

I submitted a patch for speeding up document load. Specially for objects with a
lot of modifiers.

http://sourceforge.net/tracker/index.php?func=detail&aid=2101862&group_id=11113&atid=311113

I haven't dig too deep
On k3dsdk/serialization_xml.cpp file line 1401
Document.pipeline().set_dependencies(dependencies);
Seems to evaluate properties on every dependency update.
This slows down considerably on document loading
Thus i send a patch for this case with a method for this special case.
Document.pipeline().set_dependencies_load_document(dependencies);

Copying and commenting out the following lines in a new
set_dependencies_load_document() method seems to solve the problem.

// If we're recording undo/redo data, keep track of the original state ...
if(state_recorder && state_recorder->current_change_set())
state_recorder->current_change_set()->record_old_state(new
set_dependencies_container(*this, old_dependencies));

// Notify observers that the pipeline has changed ...
changed_signal.emit(Dependencies);

// Synthesize change notifications for every property whose parent was set ...
for(dependencies_t::iterator dependency = Dependencies.begin(); dependency !=
Dependencies.end(); ++dependency)
dependency->first->property_changed_signal().emit(Hint);

I don't know if this change has any other side effect.

Cheers!
Joaquín

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

Re: Quiet out there

2008-09-10T00:24:23Z

Bart Janssens wrote:

> On 9/7/08, Timothy M. Shead <tshead@...> wrote:
>
>> * I've been making good progress on generic mesh primitives:
>>
>> http://www.k-3d.org/wiki/Generic_Mesh_Primitives#Implementation
>>
>>
>
> Looks very promising! I assume that once this is complete, the current
> array meshes will have to co-exist with the generic mesh primitives
> until all other plugins have been converted? At least it should be a
> whole lot more straightforward than the (on-going) legacy::mesh to
> array-mesh conversion :

That's the idea.

Cheers,
Tim

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development