Re: [Paraview] Is USE_EXTERNAL_VTK supported with ParaView 5.1.2?

[corinna reuter]

You'll save many people's frustration by including VTKTargets.cmake in the
"make install" target, thank you.
I found a workaround is to copy VTKTargets.cmake from ParaView's build
directory to the same place as the installed VTKConfig.cmake, and to
replace all occurrences of the build directory by their installation
counterparts.

Thank you
Corinna

On Mon, Aug 29, 2016 at 2:42 PM, Utkarsh Ayachit <
utkarsh.ayac...@kitware.com> wrote:

> > Would it be easy for you to make the ParaView installation directory
> usable
> > as VTK_DIR by adding the required cmake files? Or is there a more
> difficult
> > problem underneath? The missing VTKTargets.cmake has been discussed
> several
> > times in the past with no real solution.
>
> We can certainly support that use-case. I've reported an issue:
> https://gitlab.kitware.com/paraview/paraview/issues/16865
>
> Utkarsh
>
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Search the list archives at: http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview

Re: [Paraview] Is USE_EXTERNAL_VTK supported with ParaView 5.1.2?

[corinna reuter]

Thank you, I will try if that works. Now that you mention it, I think I
remember that I have been successful in a similar situation quite some time
ago by using the build directory.
However, the problem with this is that I would have to always copy around
the complete build directory between production servers. At the moment I
only install the installation directory on those servers. (The much smaller
installation directory helps to reduce the startup time significantly.) In
the installation directory, "VTKTargets.cmake" is missing, so that other
projects cannot use this as VTK_DIR.
Would it be easy for you to make the ParaView installation directory usable
as VTK_DIR by adding the required cmake files? Or is there a more difficult
problem underneath? The missing VTKTargets.cmake has been discussed several
times in the past with no real solution.

Thank you
Corinna

On Sat, Aug 27, 2016 at 12:27 PM, Utkarsh Ayachit <
utkarsh.ayac...@kitware.com> wrote:

> > For this purpose I configure ParaView with the
> > -DUSE_EXTERNAL_VTK=ON option, but I get a bunch of errors.
> > Before I attach lengthy error logs, I'd like to ask whether this can
> work at
> > all?
>
> No, it's not supported and there are no indications that it will any
> time soon unless someone from the community takes it up.
>
> > The background of my request is that I need to develop a ParaView plugin
> > which links a software tool that depends on VTK by itself. I couldn't
> find a
> > possibility to link this tool against the VTK which is included in
> ParaView.
>
> You should be able to set your VTK_DIR to ParaViewBuildDir/VTK.
>
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Search the list archives at: http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview

[Paraview] Is USE_EXTERNAL_VTK supported with ParaView 5.1.2?

[corinna reuter]

Hello,
I am trying to first build and install VTK-7.0.0 and then compile ParaView
5.1.2, which should use the installed VTK-7.0.0 instead of building its own
VTK version. For this purpose I configure ParaView with the
-DUSE_EXTERNAL_VTK=ON option, but I get a bunch of errors.

Before I attach lengthy error logs, I'd like to ask whether this can work
at all? It is not clear from older bug reports, whether USE_EXTERNAL_VTK is
finally supported or whether it is in some buggy, experimental stage since
3 years. If it is supported, are the 5.1.2 and 7.0.0 versions compatible,
or should I try a different combination?

If somebody can confirm that this should work in principle, I'll try to
solve as many compilation problems as possible before I send the remaining
error output for further discussion.

The background of my request is that I need to develop a ParaView plugin
which links a software tool that depends on VTK by itself. I couldn't find
a possibility to link this tool against the VTK which is included in
ParaView. Therefore I am trying to first build VTK, and then link both
ParaView and the software tool with the compiled VTK. I hope this makes
sense.

Thank you
Corinna
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Search the list archives at: http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview

Re: [Paraview] Feasibility of and documentation on implementing a Paraview Client.

[corinna reuter]

Sorry, my Paraview experience is too limited and my Unreal knowledge is
zero, so that I won't be able to give you anything else than some ideas.
First of all, Paraview cannot compete with game engines in terms of
rendering speed, because the data structures are more complicated.
Still, if you write your own "Paraview client", you'll end up with the
original paraview application reduced to only the 3d view widget. You could
suppress its output and grab the image data.
I guess you have already considered fixing the OpenGL setup. If you are
running on a headless cluster, you could install the osmesa implementation.
But without a thorough understanding of how you are interfacing with the
Unreal engine in detail and what data type and sizes and frame rates you
expect, I cannot suggest anything else than following the wiki on custom
Paraview applications. I am sure, others on this list will help.

Corinna

On Tue, Aug 16, 2016 at 2:02 PM, Magnus Elden 
wrote:

> Hi,
>
>
>
> I think you should explain why you don't want to use the paraview
> application as your client.
>
>
>
> I am using the Unreal Engine 4 for my project and I want to use the power
> of Paraview to read and render scientific volume data. I tried writing a
> small program using the VTK library, but I was unable to get it to work due
> to the program complaining about not being able to find/create an OpenGL
> context. As such I am forced to separate the rendering and the Unreal
> program. Having the client and the servers separated is something that
> Paraview supports natively and as such I do not imagine that it should be
> more difficult than to implement a code only client that my UE4 program
> uses to interface with the Paraview servers.
>
>
>
> As you can see, I am unable to use the paraview application because I will
> be using another application. Since UE4 has no real support for volume data
> or scientific visualization I am left between a choice of implementing
> everything that Paraview already has or writing a new client that handles
> the translation between my UE4 program and the Paraview server.
>
> The former option will force me to spend a lot of time implementing an
> inferior version of parts of Paraview and the latter require me to create a
> client that is able to communicate properly with the servers. I would argue
> that the latter is less work if good documentation and specifications are
> to be found. Further, I would think that the project can scale much more
> effectively if I am able to utilize the Paraview servers.
>
>
>
> I hope to feed the paraview servers with user made updates and then feed
> the resulting render back to the UE4 application in order to present it to
> the user.
>
>
>
> Yours,
>
> Magnus Elden
>
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Search the list archives at: http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview

Re: [Paraview] Feasibility of and documentation on implementing a Paraview Client.

[corinna reuter]

On Mon, Aug 15, 2016 at 9:39 PM, Magnus Elden 
wrote:

> Hi,
>
>
>
> For various reasons I aim to replace the standard client of paraview and
> use the data and rendering servers as is. As such I need to implement the
> client that communicates with the servers.
>

You should elaborate on the "various reasons". All that you described is
done by the paraview application, connected to a remove pvserver. You can
extend or reduce paraview's functionality (
http://www.paraview.org/Wiki/Writing_Custom_Applications).


> I have been unable to find any documentation on this and I would
> appreciate pointers to any documentation of protocols and APIs that are
> used.
>

Apart from the wiki, you can only search the web for discussions of similar
topics or read the Paraview Guide (book).


>
>
> The reason for why I am doing this is because I am trying to utilize the
> power of Paraview and VTK inside another system. The problem is that my VTK
> program that acts as a light renderer does not work inside the system
> because it can not obtain an OpenGL context. As such I wish to remove the
> rendering from the system and instead communicate like the design already
> permits.
>

I think you should explain why you don't want to use the paraview
application as your client.

Corinna
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Search the list archives at: http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview

Re: [Paraview] Custom widget integration in Paraview

[corinna reuter]

On Fri, Aug 12, 2016 at 5:55 PM, Cory Quammen 
wrote:

> Corinna,
>
> Does this example help?
>
> paraview/Examples/Plugins/PropertyWidgets
>

Yes, this looks like the information I need. Though I haven't found the
time to apply it to my problem, but I expect no difficulties.
Thank you!
Corinna
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Search the list archives at: http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview

[Paraview] Custom widget integration in Paraview

[corinna reuter]

Hello,
I am looking for a possibility how to integrate a custom widget into the
ParaView UI. (It is basically a text widget displaying properties of a
custom filter. In addition to the standard property widgets I have to
modify the widget's notification system. But these are implementation
details, and all I need to know from you at the moment, how I can add a
simple textfield into the UI.)

I found the ADD_3DWIDGET macro in ParaviewPlugins.cmake and a few
unanswered threads discussing whether this macro serves the purpose. It
seems like it does. But there is no documentation or example how to proceed.

Does somebody have a code snippet or a link to a description that explains
how to define custom widgets either in a plugin or hardcoded into Paraview?
Either using ADD_3DWIDGET or any other solution?

Thank you
Corinna
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Search the list archives at: http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview

Re: [Paraview] custom filter doesn't update property in the Paraview UI

[corinna reuter]

Hi Cory,
that's fine, I have some experience developing Qt UIs. But there is one
difficulty with ParaView: I couldn't find any examples or documentation how
to extend the existing widgets? Where do I plug in the additional elements,
is it possible in some xml plugin description, or do I have to hard code it
into the ParaView sources?

The manual update is fine, although not perfect. Ideally I would like to
let the user choose once whether update is automatic or manual. Since the
manual update button of the Threshold filter changes its color when the
filter input is modified, I assume that I should be able to use the event
that triggers the color change to trigger the update of the displayed
property instead.

Thank you
Corinna

On Mon, Aug 8, 2016 at 8:29 PM, Cory Quammen <cory.quam...@kitware.com>
wrote:

> Hi Corinna,
>
> Utkarsh addressed the updating of properties in his earlier response
> about the directionality of property changes in ParaView:
>
> "
> 2. ParaView's properties are not bi-directional. They are only
> intended to be set on the UI side and pushed to the VTK side. There
> are of course exceptions to this, and you can add what's called a
> "information_only" property that instead reads a value from the VTK
> side and provides it on the "client" side, but for such properties,
> often there's no UI. They are instead, used to define the "domain" for
> the property which controls things like range for the slider, for
> example.
> "
>
> So, in short, ParaView is not really designed to do this. However, it
> may be possible to define a custom panel for your NPoints property
> that, when clicked, updates the property value to the current scalar
> range. For example, there is a custom panel widget in the Threshold
> filter that lets you update the ranges based on the range of the
> currently selected scalar array. You would need this kind of custom
> panel widget for your NPoints property. The only catch is you would
> need to update the value manually, or perhaps you could use a QTimer
> to periodically refresh the value.
>
> HTH,
> Cory
>
> On Mon, Aug 8, 2016 at 2:14 PM, corinna reuter
> <corinnareute...@gmail.com> wrote:
> > Dear Cory,
> > I wonder, if you still have an idea how to implement a filter that
> > initializes its properties from upstream objects and updates these
> > properties when those objects change.
> > Isn't the whole ParaView designed such that changes are propagated along
> the
> > pipeline? Why is it so complicated to render propagated changes in the
> > filter parameters visible?
> >
> >
> > Thank you
> > Corinna
> >
> >
> > On Tue, Aug 2, 2016 at 4:55 PM, corinna reuter <
> corinnareute...@gmail.com>
> > wrote:
> >>
> >> Dear Cory, dear all,
> >> sorry for not getting back to this topic earlier. Your solution works
> >> perfectly for the described problem. When I create e.g. a SphereSource
> and
> >> attach the TestProperty filter with its NPoints property, this property
> is
> >> initialized with the correct number of points in the sphere.
> >>
> >> However, I have another requirement which is that the property should
> also
> >> be updated, when the input changes. I thought the same mechanism for
> >> initializing the property would also take care of updating the property
> upon
> >> modified input.
> >>
> >> If I modify the SphereSource from above by changing the ThetaResolution
> or
> >> PhiResolution, then vtkTestProperty::RequestInformation is called as
> >> expected, and the included print statements show the new number of
> points,
> >> but the NPoints property won't reflect this new value. Do you know of
> >> another short trick, how to fix this? You mentioned that you found the
> need
> >> for a domain element in the ParaView source code. Where exactly? If I
> >> understand what prevented the property initialization, I might be able
> to
> >> find out what prevents the update.
> >>
> >> The complete filter sources, updated to include your corrections, are
> >> attached.
> >>
> >> Thank you
> >> Corinna
> >>
> >> On Fri, Jul 15, 2016 at 6:02 PM, Cory Quammen <cory.quam...@kitware.com
> >
> >> wrote:
> >>>
> >>> Corinna,
> >>>
> >>> Looking at the ParaView source code, I believe that vector properties
> >>> with information properties must have a domain associated with them.
> >>> In your case, you can add an IntRangeDomain with only a minimum
> >&

Re: [Paraview] custom filter doesn't update property in the Paraview UI

[corinna reuter]

Dear Cory,
I wonder, if you still have an idea how to implement a filter that
initializes its properties from upstream objects and updates these
properties when those objects change.
Isn't the whole ParaView designed such that changes are propagated along
the pipeline? Why is it so complicated to render propagated changes in the
filter parameters visible?


Thank you
Corinna


On Tue, Aug 2, 2016 at 4:55 PM, corinna reuter <corinnareute...@gmail.com>
wrote:

> Dear Cory, dear all,
> sorry for not getting back to this topic earlier. Your solution works
> perfectly for the described problem. When I create e.g. a SphereSource and
> attach the TestProperty filter with its NPoints property, this property is
> initialized with the correct number of points in the sphere.
>
> However, I have another requirement which is that the property should also
> be updated, when the input changes. I thought the same mechanism for
> initializing the property would also take care of updating the property
> upon modified input.
>
> If I modify the SphereSource from above by changing the ThetaResolution or
> PhiResolution, then vtkTestProperty::RequestInformation is called as
> expected, and the included print statements show the new number of points,
> but the NPoints property won't reflect this new value. Do you know of
> another short trick, how to fix this? You mentioned that you found the need
> for a domain element in the ParaView source code. Where exactly? If I
> understand what prevented the property initialization, I might be able to
> find out what prevents the update.
>
> The complete filter sources, updated to include your corrections, are
> attached.
>
> Thank you
> Corinna
>
> On Fri, Jul 15, 2016 at 6:02 PM, Cory Quammen <cory.quam...@kitware.com>
> wrote:
>
>> Corinna,
>>
>> Looking at the ParaView source code, I believe that vector properties
>> with information properties must have a domain associated with them.
>> In your case, you can add an IntRangeDomain with only a minimum
>> defined:
>>
>> > command="SetNPoints"
>> information_property="NPointsInfo"
>> number_of_elements="1"
>> default_values="42">
>> 
>> 
>>
>> 0 seems like a reasonable minimum in your case.
>>
>> Keep your information_only property as it is. You'll need to change
>> where your NPoints is set, though - moving this to the
>> RequestInformation() member function should work.
>>
>> Cory
>>
>>
>> On Fri, Jul 15, 2016 at 3:46 AM, corinna reuter
>> <corinnareute...@gmail.com> wrote:
>> > Cory, thanks for mentioning the TIFFReader proxy. I tried to modify my
>> > previously attached code such that the NPoints property links to a
>> > NPointsInfo property. Plus, I have removed the call
>> > "this->SetNPoints(nset);" inside vtkTestProperty::RequestData as
>> suggested
>> > by Utkarsh. The NPoints property is still showing the default "42",
>> though,
>> > and not initialized or updated from the number of points of the input
>> > dataset. I am obviously missing another important step in addition to
>> > changing the xml definitions, but which one?
>> >
>> > Here is the detailed modification of TestProperty.xml compared to the
>> > earlier attachment:
>> > replace
>> > "> > command="SetNPoints"
>> > number_of_elements="1"
>> > default_values="42">
>> >   "
>> > with:
>> > "> > command="SetNPoints"
>> > information_property="NPointsInfo"
>> > number_of_elements="1"
>> > default_values="42">
>> >   
>> >   > > command="GetNPoints"
>> > information_only="1">
>> > 
>> >   "
>> >
>> > Thank you
>> > Corinna
>> >
>> >
>> > On Thu, Jul 14, 2016 at 5:24 PM, Cory Quammen <cory.quam...@kitware.com
>> >
>> > wrote:
>> >>
>> >> Corinna,
>> >>
>> >> Take a look at the TIFFReader SourceProxy in
>> >> ParaViewCore/ServerManager/SMApplication/Resources/readers.xml. It has
>> >> a CustomDataSpacing property th

Re: [Paraview] custom filter doesn't update property in the Paraview UI

[corinna reuter]

Dear Cory, dear all,
sorry for not getting back to this topic earlier. Your solution works
perfectly for the described problem. When I create e.g. a SphereSource and
attach the TestProperty filter with its NPoints property, this property is
initialized with the correct number of points in the sphere.

However, I have another requirement which is that the property should also
be updated, when the input changes. I thought the same mechanism for
initializing the property would also take care of updating the property
upon modified input.

If I modify the SphereSource from above by changing the ThetaResolution or
PhiResolution, then vtkTestProperty::RequestInformation is called as
expected, and the included print statements show the new number of points,
but the NPoints property won't reflect this new value. Do you know of
another short trick, how to fix this? You mentioned that you found the need
for a domain element in the ParaView source code. Where exactly? If I
understand what prevented the property initialization, I might be able to
find out what prevents the update.

The complete filter sources, updated to include your corrections, are
attached.

Thank you
Corinna

On Fri, Jul 15, 2016 at 6:02 PM, Cory Quammen <cory.quam...@kitware.com>
wrote:

> Corinna,
>
> Looking at the ParaView source code, I believe that vector properties
> with information properties must have a domain associated with them.
> In your case, you can add an IntRangeDomain with only a minimum
> defined:
>
>  command="SetNPoints"
> information_property="NPointsInfo"
> number_of_elements="1"
> default_values="42">
> 
> 
>
> 0 seems like a reasonable minimum in your case.
>
> Keep your information_only property as it is. You'll need to change
> where your NPoints is set, though - moving this to the
> RequestInformation() member function should work.
>
> Cory
>
>
> On Fri, Jul 15, 2016 at 3:46 AM, corinna reuter
> <corinnareute...@gmail.com> wrote:
> > Cory, thanks for mentioning the TIFFReader proxy. I tried to modify my
> > previously attached code such that the NPoints property links to a
> > NPointsInfo property. Plus, I have removed the call
> > "this->SetNPoints(nset);" inside vtkTestProperty::RequestData as
> suggested
> > by Utkarsh. The NPoints property is still showing the default "42",
> though,
> > and not initialized or updated from the number of points of the input
> > dataset. I am obviously missing another important step in addition to
> > changing the xml definitions, but which one?
> >
> > Here is the detailed modification of TestProperty.xml compared to the
> > earlier attachment:
> > replace
> > " > command="SetNPoints"
> > number_of_elements="1"
> > default_values="42">
> >   "
> > with:
> > " > command="SetNPoints"
> > information_property="NPointsInfo"
> > number_of_elements="1"
> > default_values="42">
> >   
> >> command="GetNPoints"
> > information_only="1">
> > 
> >   "
> >
> > Thank you
> > Corinna
> >
> >
> > On Thu, Jul 14, 2016 at 5:24 PM, Cory Quammen <cory.quam...@kitware.com>
> > wrote:
> >>
> >> Corinna,
> >>
> >> Take a look at the TIFFReader SourceProxy in
> >> ParaViewCore/ServerManager/SMApplication/Resources/readers.xml. It has
> >> a CustomDataSpacing property that enables you to set voxel spacing,
> >> and has an associated information property CustomDataSpacingInfo that
> >> should initially populate the spacing property when the reader is
> >> created.
> >>
> >> HTH,
> >> Cory
> >>
> >> On Thu, Jul 14, 2016 at 11:17 AM, Utkarsh Ayachit
> >> <utkarsh.ayac...@kitware.com> wrote:
> >> >> I'll check the meaning of MTime.
> >> >
> >> > See the implementation of vtkSetMacro() in vtkSetGet.h. Every time the
> >> > value changes, it call this->Modified() which updates the MTime. Since
> >> > VTK is a demand-driven pipeline, changes to MTime is what tells the
> >> > pipeline that the algorithm has been modified and needs reexecution.
> >> > If one modifies the MTime

Re: [Paraview] custom filter doesn't update property in the Paraview UI

[corinna reuter]

Cory, thanks for mentioning the TIFFReader proxy. I tried to modify my
previously attached code such that the NPoints property links to a
NPointsInfo property. Plus, I have removed the call
"this->SetNPoints(nset);" inside vtkTestProperty::RequestData as suggested
by Utkarsh. The NPoints property is still showing the default "42", though,
and not initialized or updated from the number of points of the input
dataset. I am obviously missing another important step in addition to
changing the xml definitions, but which one?

Here is the detailed modification of TestProperty.xml compared to the
earlier attachment:
replace
"
  "
with:
"
  
  

  "

Thank you
Corinna


On Thu, Jul 14, 2016 at 5:24 PM, Cory Quammen 
wrote:

> Corinna,
>
> Take a look at the TIFFReader SourceProxy in
> ParaViewCore/ServerManager/SMApplication/Resources/readers.xml. It has
> a CustomDataSpacing property that enables you to set voxel spacing,
> and has an associated information property CustomDataSpacingInfo that
> should initially populate the spacing property when the reader is
> created.
>
> HTH,
> Cory
>
> On Thu, Jul 14, 2016 at 11:17 AM, Utkarsh Ayachit
>  wrote:
> >> I'll check the meaning of MTime.
> >
> > See the implementation of vtkSetMacro() in vtkSetGet.h. Every time the
> > value changes, it call this->Modified() which updates the MTime. Since
> > VTK is a demand-driven pipeline, changes to MTime is what tells the
> > pipeline that the algorithm has been modified and needs reexecution.
> > If one modifies the MTime in RequestData() (the method called by the
> > pipeline to process the input to produce output), the pipeline thinks
> > the filter is always dirty and that can cause unexpected side effects.
> >
> >
> >> I'd need something similar to an image resize filter. The output size
> should
> >> be identical to the input size, when the filter is first attached to the
> >> input. The output size should be shown in the UI and the user should be
> able
> >> to enter a different output size.
> >
> > This is possible. Let me see if I can find you a simple example for
> > the same. I'll get back to you.
> > ___
> > Powered by www.kitware.com
> >
> > Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
> >
> > Please keep messages on-topic and check the ParaView Wiki at:
> http://paraview.org/Wiki/ParaView
> >
> > Search the list archives at: http://markmail.org/search/?q=ParaView
> >
> > Follow this link to subscribe/unsubscribe:
> > http://public.kitware.com/mailman/listinfo/paraview
>
>
>
> --
> Cory Quammen
> R Engineer
> Kitware, Inc.
>
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Search the list archives at: http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview

Re: [Paraview] custom filter doesn't update property in the Paraview UI

[corinna reuter]

On Thu, Jul 14, 2016 at 4:40 PM, Utkarsh Ayachit <
utkarsh.ayac...@kitware.com> wrote:

> There are a couple of things wrong with your plugin :).
>

I was expecting this, because I am only starting to learn plugin
programming right now :)


>
> 1. A RequestData() implementation should never change MTime of the
> filter. You code call SetNPoints() which changes MTime.
>

I'll check the meaning of MTime.

2. ParaView's properties are not bi-directional. They are only
> intended to be set on the UI side and pushed to the VTK side.


I was also assuming this, but I thought there are a few filters that update
their properties when some inputs upstream in the pipeline change. My
memory can be wrong, though.


> There
> are of course exceptions to this, and you can add what's called a
> "information_only" property that instead reads a value from the VTK
> side and provides it on the "client" side, but for such properties,
> often there's no UI. They are instead, used to define the "domain" for
> the property which controls things like range for the slider, for
> example.
>

I am not sure, whether I understand all details of your reply.
In summary, is it possible to update UI properties of a filter, when its
input changes? By no means?

I'd need something similar to an image resize filter. The output size
should be identical to the input size, when the filter is first attached to
the input. The output size should be shown in the UI and the user should be
able to enter a different output size. Is this impossible? Not even with
some more complicated workaround?

Thank you
Corinna



>
>
>
> On Thu, Jul 14, 2016 at 10:34 AM, corinna reuter
> <corinnareute...@gmail.com> wrote:
> > Hello,
> > I would like to work on a filter that updates some of its properties from
> > the filter input. As a demonstration, I attach a (quite minimal)
> > "vtkTestProperty" filter algorithm, which takes vtkPolyData input, and
> which
> > is supposed to copy input->GetNumberOfPoints() into the filter's
> "NPoints"
> > property which is exposed to the user interface.
> >
> > Inside vtkTestProperty::RequestData, I execute:
> > int nset=input->GetNumberOfPoints();
> > this->SetNPoints(nset);
> > which, is correctly setting the NPoints value. Additional print
> statements
> > in the code confirm, that NPoints is actually updated, but the new value
> > won't be shown in the Paraview UI.
> >
> > To reproduce the behaviour, the attachment is ready to compile with
> Paraview
> > 5.0.1. Just load the compiled plugin in Paraview, create a "SphereSource"
> > and append the "ATestPropery" filter to the SphereSource. The console
> output
> > will show that NPoints is set to 50 (Number of Points of the Sphere), but
> > the UI still shows 42 (the default value from the Servermanager xml).
> >
> > What am I doing wrong?
> >
> > Thank you
> > Corinna
> >
> > ___
> > Powered by www.kitware.com
> >
> > Visit other Kitware open-source projects at
> > http://www.kitware.com/opensource/opensource.html
> >
> > Please keep messages on-topic and check the ParaView Wiki at:
> > http://paraview.org/Wiki/ParaView
> >
> > Search the list archives at: http://markmail.org/search/?q=ParaView
> >
> > Follow this link to subscribe/unsubscribe:
> > http://public.kitware.com/mailman/listinfo/paraview
> >
>
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Search the list archives at: http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview

[Paraview] custom filter doesn't update property in the Paraview UI

[corinna reuter]

Hello,
I would like to work on a filter that updates some of its properties from
the filter input. As a demonstration, I attach a (quite minimal)
"vtkTestProperty" filter algorithm, which takes vtkPolyData input, and
which is supposed to copy input->GetNumberOfPoints() into the filter's
"NPoints" property which is exposed to the user interface.

Inside vtkTestProperty::RequestData, I execute:
int nset=input->GetNumberOfPoints();
this->SetNPoints(nset);
which, is correctly setting the NPoints value. Additional print statements
in the code confirm, that NPoints is actually updated, but the new value
won't be shown in the Paraview UI.

To reproduce the behaviour, the attachment is ready to compile with
Paraview 5.0.1. Just load the compiled plugin in Paraview, create a
"SphereSource" and append the "ATestPropery" filter to the SphereSource.
The console output will show that NPoints is set to 50 (Number of Points of
the Sphere), but the UI still shows 42 (the default value from the
Servermanager xml).

What am I doing wrong?

Thank you
Corinna


TestProperty.tgz
Description: GNU Zip compressed data
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Search the list archives at: http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview

Re: [Paraview] plugin error: "Input for algorithm vtkPVPostFilter is of type ..., but a vtkPolyData is required"

[corinna reuter]

Thank you! After recompiling Paraview with that patch, the problem seems to
be solved.
Will the patch be automatically merged into the development version? And in
which future Paraview release can we expect to find it? (I don't understand
the details of the build robot's messages)

Corinna

On Wed, Jul 13, 2016 at 5:25 PM, Cory Quammen <cory.quam...@kitware.com>
wrote:

> Corinna,
>
> It seems that your use case is not currently supported. T.J. (CC'ed)
> is working on a bug fix in ParaView that should enable what you are
> doing.
>
> https://gitlab.kitware.com/paraview/paraview/merge_requests/404
>
> He can fill you in with more details if you need.
>
> Thanks,
> Cory
>
> On Wed, Jul 13, 2016 at 2:13 AM, corinna reuter
> <corinnareute...@gmail.com> wrote:
> > Hi Cory,
> > can you still help me to find the problem, why the custom vtkTest source
> > cannot be produced inside paraview, but fails with a vtkPVPostFilter
> message
> > (see subject)? A complete test was attached to my previous email.
> >
> > Why does vtkPVPostFilter expect vtkPolyData input, although both in the
> > Servermanager xml and in FillOutputPortInformation the type is correctly
> > given as vtkTest? Is there another place, where I need to define the
> output
> > type?
> >
> > I also tried to make use of the vtkTableReader, but I can't seem to make
> it
> > work, because it tries to read point data, cell data etc., which don't
> exist
> > a priori in my use case.
> >
> > Corinna
> >
> >
> > On Fri, Jul 8, 2016 at 10:03 AM, corinna reuter <
> corinnareute...@gmail.com>
> > wrote:
> >>
> >> Hi Cory,
> >> I attach the complete sources for compiling and testing the described
> >> problem.
> >> (Tested with Paraview 5.0.1 compiled from sources.)
> >>
> >> Corinna
> >>
> >> On Thu, Jul 7, 2016 at 7:31 PM, corinna reuter <
> corinnareute...@gmail.com>
> >> wrote:
> >>>
> >>> Yes, FillOutputPortInformation already looks exactly as you say.
> >>>
> >>> I gave the link to the complete example code in my initial question:
> >>> http://www.vtk.org/Wiki/VTK_Examples_vtkAlgorithm_Source
> >>> I changed a few lines to make it compile with Paraview 5.0.1 (see the
> >>> comments in the CMakeLists.txt with my initial question). In addition,
> I can
> >>> send to you tomorrow a small source archive ready for compilation.
> >>>
> >>> I'll also check vtkTable to see if it solves my problem. I thought of
> >>> just deriving the vtkTest class from vtkDataSet and returning a fake,
> empty
> >>> dataset while keeping the information from which the actual vtkDataSet
> will
> >>> be processed downstream in the pipeline. Do you think that this
> workaround
> >>> is feasible, or what other problems should I excpect?
> >>>
> >>> Thank you
> >>> Corinna
> >>>
> >>>
> >>> On Thu, Jul 7, 2016 at 3:03 PM, Cory Quammen <cory.quam...@kitware.com
> >
> >>> wrote:
> >>>>
> >>>> Hi Corinna,
> >>>>
> >>>> Does vtkTestSource override the FillOutputPortInformation() member
> >>>> function? It should look like
> >>>>
> >>>>
> >>>>
> //
> >>>> int vtkTestSource::FillOutputPortInformation(
> >>>>   int vtkNotUsed(port), vtkInformation* info)
> >>>> {
> >>>>   // now add our info
> >>>>   info->Set(vtkDataObject::DATA_TYPE_NAME(), "vtkTest");
> >>>>   return 1;
> >>>> }
> >>>>
> >>>> If that doesn't fix the error, let us know.
> >>>>
> >>>> You may run into other problems with a custom data object - you may
> >>>> want to derive your vtkTest class from one of the vtkDataObject
> >>>> ParaView works with, e.g., vtkTable.
> >>>>
> >>>> Thanks,
> >>>> Cory
> >>>>
> >>>> On Thu, Jul 7, 2016 at 1:44 AM, corinna reuter
> >>>> <corinnareute...@gmail.com> wrote:
> >>>> > Doesn't anybody know, why a vtkAlgorithm which produces a custom
> class
> >>>> > (vtkTest) that is not derived from vtkDataSet show the error "...
> but
> >>>> > a
> >>>> > vtkPolyData is requ

Re: [Paraview] plugin error: "Input for algorithm vtkPVPostFilter is of type ..., but a vtkPolyData is required"

[corinna reuter]

Hi Cory,
can you still help me to find the problem, why the custom vtkTest source
cannot be produced inside paraview, but fails with a vtkPVPostFilter
message (see subject)? A complete test was attached to my previous email.

Why does vtkPVPostFilter expect vtkPolyData input, although both in the
Servermanager xml and in FillOutputPortInformation the type is correctly
given as vtkTest? Is there another place, where I need to define the output
type?

I also tried to make use of the vtkTableReader, but I can't seem to make it
work, because it tries to read point data, cell data etc., which don't
exist a priori in my use case.

Corinna

On Fri, Jul 8, 2016 at 10:03 AM, corinna reuter <corinnareute...@gmail.com>
wrote:

> Hi Cory,
> I attach the complete sources for compiling and testing the described
> problem.
> (Tested with Paraview 5.0.1 compiled from sources.)
>
> Corinna
>
> On Thu, Jul 7, 2016 at 7:31 PM, corinna reuter <corinnareute...@gmail.com>
> wrote:
>
>> Yes, FillOutputPortInformation already looks exactly as you say.
>>
>> I gave the link to the complete example code in my initial question:
>> http://www.vtk.org/Wiki/VTK_Examples_vtkAlgorithm_Source
>> I changed a few lines to make it compile with Paraview 5.0.1 (see the
>> comments in the CMakeLists.txt with my initial question). In addition, I
>> can send to you tomorrow a small source archive ready for compilation.
>>
>> I'll also check vtkTable to see if it solves my problem. I thought of
>> just deriving the vtkTest class from vtkDataSet and returning a fake, empty
>> dataset while keeping the information from which the actual vtkDataSet will
>> be processed downstream in the pipeline. Do you think that this workaround
>> is feasible, or what other problems should I excpect?
>>
>> Thank you
>> Corinna
>>
>>
>> On Thu, Jul 7, 2016 at 3:03 PM, Cory Quammen <cory.quam...@kitware.com>
>> wrote:
>>
>>> Hi Corinna,
>>>
>>> Does vtkTestSource override the FillOutputPortInformation() member
>>> function? It should look like
>>>
>>>
>>> //
>>> int vtkTestSource::FillOutputPortInformation(
>>>   int vtkNotUsed(port), vtkInformation* info)
>>> {
>>>   // now add our info
>>>   info->Set(vtkDataObject::DATA_TYPE_NAME(), "vtkTest");
>>>   return 1;
>>> }
>>>
>>> If that doesn't fix the error, let us know.
>>>
>>> You may run into other problems with a custom data object - you may
>>> want to derive your vtkTest class from one of the vtkDataObject
>>> ParaView works with, e.g., vtkTable.
>>>
>>> Thanks,
>>> Cory
>>>
>>> On Thu, Jul 7, 2016 at 1:44 AM, corinna reuter
>>> <corinnareute...@gmail.com> wrote:
>>> > Doesn't anybody know, why a vtkAlgorithm which produces a custom class
>>> > (vtkTest) that is not derived from vtkDataSet show the error "... but a
>>> > vtkPolyData is required"?
>>> >
>>> > A side remark, previously I wrote that the custom class "vtkTest" is
>>> neither
>>> > derived from vtkDataSet nor from vtkDataObject. Actually it is derived
>>> from
>>> > vtkDataObject.
>>> >
>>> > If my question doesn't make sense, please let me know. Maybe it is
>>> against
>>> > every paraview design, so that I'll have to find a completely different
>>> > solution. But it becomes more and more urgent for me to include a data
>>> > source (mySQL database) of custom classes (not vtkDataSets) in the
>>> paraview
>>> > tree.
>>> >
>>> > Thank you
>>> > Corinna
>>> >
>>> > On Mon, Jul 4, 2016 at 3:09 PM, corinna reuter <
>>> corinnareute...@gmail.com>
>>> > wrote:
>>> >>
>>> >> To anybody experienced in plugin programming:
>>> >>
>>> >> I am trying to include items in the paraview tree, which are not
>>> derived
>>> >> from vtkDataSet or vtkDataObject. For example, a mySQL database
>>> should be
>>> >> allowed as a data source, which delivers data in an intermediate
>>> format.
>>> >> Next I'd implement filters which take the intermediate format and
>>> produce
>>> >> vtkDataSets as output.
>>> >>
>>> >> Is it possible to use such "abstract" (not vtkDataSet) objects in the

Re: [Paraview] plugin error: "Input for algorithm vtkPVPostFilter is of type ..., but a vtkPolyData is required"

[corinna reuter]

Hi Cory,
I attach the complete sources for compiling and testing the described
problem.
(Tested with Paraview 5.0.1 compiled from sources.)

Corinna

On Thu, Jul 7, 2016 at 7:31 PM, corinna reuter <corinnareute...@gmail.com>
wrote:

> Yes, FillOutputPortInformation already looks exactly as you say.
>
> I gave the link to the complete example code in my initial question:
> http://www.vtk.org/Wiki/VTK_Examples_vtkAlgorithm_Source
> I changed a few lines to make it compile with Paraview 5.0.1 (see the
> comments in the CMakeLists.txt with my initial question). In addition, I
> can send to you tomorrow a small source archive ready for compilation.
>
> I'll also check vtkTable to see if it solves my problem. I thought of just
> deriving the vtkTest class from vtkDataSet and returning a fake, empty
> dataset while keeping the information from which the actual vtkDataSet will
> be processed downstream in the pipeline. Do you think that this workaround
> is feasible, or what other problems should I excpect?
>
> Thank you
> Corinna
>
>
> On Thu, Jul 7, 2016 at 3:03 PM, Cory Quammen <cory.quam...@kitware.com>
> wrote:
>
>> Hi Corinna,
>>
>> Does vtkTestSource override the FillOutputPortInformation() member
>> function? It should look like
>>
>>
>> //
>> int vtkTestSource::FillOutputPortInformation(
>>   int vtkNotUsed(port), vtkInformation* info)
>> {
>>   // now add our info
>>   info->Set(vtkDataObject::DATA_TYPE_NAME(), "vtkTest");
>>   return 1;
>> }
>>
>> If that doesn't fix the error, let us know.
>>
>> You may run into other problems with a custom data object - you may
>> want to derive your vtkTest class from one of the vtkDataObject
>> ParaView works with, e.g., vtkTable.
>>
>> Thanks,
>> Cory
>>
>> On Thu, Jul 7, 2016 at 1:44 AM, corinna reuter
>> <corinnareute...@gmail.com> wrote:
>> > Doesn't anybody know, why a vtkAlgorithm which produces a custom class
>> > (vtkTest) that is not derived from vtkDataSet show the error "... but a
>> > vtkPolyData is required"?
>> >
>> > A side remark, previously I wrote that the custom class "vtkTest" is
>> neither
>> > derived from vtkDataSet nor from vtkDataObject. Actually it is derived
>> from
>> > vtkDataObject.
>> >
>> > If my question doesn't make sense, please let me know. Maybe it is
>> against
>> > every paraview design, so that I'll have to find a completely different
>> > solution. But it becomes more and more urgent for me to include a data
>> > source (mySQL database) of custom classes (not vtkDataSets) in the
>> paraview
>> > tree.
>> >
>> > Thank you
>> > Corinna
>> >
>> > On Mon, Jul 4, 2016 at 3:09 PM, corinna reuter <
>> corinnareute...@gmail.com>
>> > wrote:
>> >>
>> >> To anybody experienced in plugin programming:
>> >>
>> >> I am trying to include items in the paraview tree, which are not
>> derived
>> >> from vtkDataSet or vtkDataObject. For example, a mySQL database should
>> be
>> >> allowed as a data source, which delivers data in an intermediate
>> format.
>> >> Next I'd implement filters which take the intermediate format and
>> produce
>> >> vtkDataSets as output.
>> >>
>> >> Is it possible to use such "abstract" (not vtkDataSet) objects in the
>> >> paraview tree as a representation of the intermediate format? I have
>> tried
>> >> to integrate the example for custom classes from
>> >> http://www.vtk.org/Wiki/VTK_Examples_vtkAlgorithm_Source into a
>> paraview
>> >> plugin (CMakeLists and servermanager xml below). When the plugin is
>> loaded
>> >> and I add a source of the new type AlgorithmSourceExample, this error
>> is
>> >> thrown:
>> >>
>> >> ERROR: In
>> >>
>> /opt/ParaView-v5.0.1/VTK/Common/ExecutionModel/vtkDemandDrivenPipeline.cxx,
>> >> line 810
>> >> vtkPVPostFilterExecutive (0x3857810): Input for connection index 0 on
>> >> input port index 0 for algorithm vtkPVPostFilter(0x3858910) is of type
>> >> vtkTest, but a vtkPolyData is required.
>> >>
>> >> What do I need to change such that the vtkTest objects, produced by
>> >> AlgorithmS

Re: [Paraview] plugin error: "Input for algorithm vtkPVPostFilter is of type ..., but a vtkPolyData is required"

[corinna reuter]

Yes, FillOutputPortInformation already looks exactly as you say.

I gave the link to the complete example code in my initial question:
http://www.vtk.org/Wiki/VTK_Examples_vtkAlgorithm_Source
I changed a few lines to make it compile with Paraview 5.0.1 (see the
comments in the CMakeLists.txt with my initial question). In addition, I
can send to you tomorrow a small source archive ready for compilation.

I'll also check vtkTable to see if it solves my problem. I thought of just
deriving the vtkTest class from vtkDataSet and returning a fake, empty
dataset while keeping the information from which the actual vtkDataSet will
be processed downstream in the pipeline. Do you think that this workaround
is feasible, or what other problems should I excpect?

Thank you
Corinna


On Thu, Jul 7, 2016 at 3:03 PM, Cory Quammen <cory.quam...@kitware.com>
wrote:

> Hi Corinna,
>
> Does vtkTestSource override the FillOutputPortInformation() member
> function? It should look like
>
>
> //
> int vtkTestSource::FillOutputPortInformation(
>   int vtkNotUsed(port), vtkInformation* info)
> {
>   // now add our info
>   info->Set(vtkDataObject::DATA_TYPE_NAME(), "vtkTest");
>   return 1;
> }
>
> If that doesn't fix the error, let us know.
>
> You may run into other problems with a custom data object - you may
> want to derive your vtkTest class from one of the vtkDataObject
> ParaView works with, e.g., vtkTable.
>
> Thanks,
> Cory
>
> On Thu, Jul 7, 2016 at 1:44 AM, corinna reuter
> <corinnareute...@gmail.com> wrote:
> > Doesn't anybody know, why a vtkAlgorithm which produces a custom class
> > (vtkTest) that is not derived from vtkDataSet show the error "... but a
> > vtkPolyData is required"?
> >
> > A side remark, previously I wrote that the custom class "vtkTest" is
> neither
> > derived from vtkDataSet nor from vtkDataObject. Actually it is derived
> from
> > vtkDataObject.
> >
> > If my question doesn't make sense, please let me know. Maybe it is
> against
> > every paraview design, so that I'll have to find a completely different
> > solution. But it becomes more and more urgent for me to include a data
> > source (mySQL database) of custom classes (not vtkDataSets) in the
> paraview
> > tree.
> >
> > Thank you
> > Corinna
> >
> > On Mon, Jul 4, 2016 at 3:09 PM, corinna reuter <
> corinnareute...@gmail.com>
> > wrote:
> >>
> >> To anybody experienced in plugin programming:
> >>
> >> I am trying to include items in the paraview tree, which are not derived
> >> from vtkDataSet or vtkDataObject. For example, a mySQL database should
> be
> >> allowed as a data source, which delivers data in an intermediate format.
> >> Next I'd implement filters which take the intermediate format and
> produce
> >> vtkDataSets as output.
> >>
> >> Is it possible to use such "abstract" (not vtkDataSet) objects in the
> >> paraview tree as a representation of the intermediate format? I have
> tried
> >> to integrate the example for custom classes from
> >> http://www.vtk.org/Wiki/VTK_Examples_vtkAlgorithm_Source into a
> paraview
> >> plugin (CMakeLists and servermanager xml below). When the plugin is
> loaded
> >> and I add a source of the new type AlgorithmSourceExample, this error is
> >> thrown:
> >>
> >> ERROR: In
> >>
> /opt/ParaView-v5.0.1/VTK/Common/ExecutionModel/vtkDemandDrivenPipeline.cxx,
> >> line 810
> >> vtkPVPostFilterExecutive (0x3857810): Input for connection index 0 on
> >> input port index 0 for algorithm vtkPVPostFilter(0x3858910) is of type
> >> vtkTest, but a vtkPolyData is required.
> >>
> >> What do I need to change such that the vtkTest objects, produced by
> >> AlgorithmSourceExample, can live in the paraview tree?
> >>
> >> Thank you
> >> Corinna
> >>
> >> CMakeLists.txt:
> >>
> >> FIND_PACKAGE(ParaView REQUIRED)
> >> INCLUDE(${PARAVIEW_USE_FILE})
> >> #changes required in
> >> http://www.vtk.org/Wiki/VTK_Examples_vtkAlgorithm_Source:
> >> # remove vtkCxxRevisionMacro calls
> >> # change vtkTypeRevisionMacro into vtkTypeMacro
> >> # remove call output->SetPipelineInformation( outInfo );
> >>  ADD_PARAVIEW_PLUGIN(
> >>AlgorithmSourceExample
> >>   "1.0"
> >>SERVER_MANAGER_XML AlgorithmSourceExample.xml
> >>SERVER_MANAGER_SOURCES vtkTestSource.cxx 

Re: [Paraview] plugin error: "Input for algorithm vtkPVPostFilter is of type ..., but a vtkPolyData is required"

[corinna reuter]

Doesn't anybody know, why a vtkAlgorithm which produces a custom class
(vtkTest) that is not derived from vtkDataSet show the error "... but a
vtkPolyData is required"?

A side remark, previously I wrote that the custom class "vtkTest" is
neither derived from vtkDataSet nor from vtkDataObject. Actually it is
derived from vtkDataObject.

If my question doesn't make sense, please let me know. Maybe it is against
every paraview design, so that I'll have to find a completely different
solution. But it becomes more and more urgent for me to include a data
source (mySQL database) of custom classes (not vtkDataSets) in the paraview
tree.

Thank you
Corinna

On Mon, Jul 4, 2016 at 3:09 PM, corinna reuter <corinnareute...@gmail.com>
wrote:

> To anybody experienced in plugin programming:
>
> I am trying to include items in the paraview tree, which are not derived
> from vtkDataSet or vtkDataObject. For example, a mySQL database should be
> allowed as a data source, which delivers data in an intermediate format.
> Next I'd implement filters which take the intermediate format and produce
> vtkDataSets as output.
>
> Is it possible to use such "abstract" (not vtkDataSet) objects in the
> paraview tree as a representation of the intermediate format? I have tried
> to integrate the example for custom classes from
> http://www.vtk.org/Wiki/VTK_Examples_vtkAlgorithm_Source into a paraview
> plugin (CMakeLists and servermanager xml below). When the plugin is loaded
> and I add a source of the new type AlgorithmSourceExample, this error is
> thrown:
>
> ERROR: In
> /opt/ParaView-v5.0.1/VTK/Common/ExecutionModel/vtkDemandDrivenPipeline.cxx,
> line 810
> vtkPVPostFilterExecutive (0x3857810): Input for connection index 0 on
> input port index 0 for algorithm vtkPVPostFilter(0x3858910) is of type
> vtkTest, but a vtkPolyData is required.
>
> What do I need to change such that the vtkTest objects, produced by
> AlgorithmSourceExample, can live in the paraview tree?
>
> Thank you
> Corinna
>
> CMakeLists.txt:
>
> FIND_PACKAGE(ParaView REQUIRED)
> INCLUDE(${PARAVIEW_USE_FILE})
> #changes required in
> http://www.vtk.org/Wiki/VTK_Examples_vtkAlgorithm_Source:
> # remove vtkCxxRevisionMacro calls
> # change vtkTypeRevisionMacro into vtkTypeMacro
> # remove call output->SetPipelineInformation( outInfo );
>  ADD_PARAVIEW_PLUGIN(
>AlgorithmSourceExample
>   "1.0"
>SERVER_MANAGER_XML AlgorithmSourceExample.xml
>SERVER_MANAGER_SOURCES vtkTestSource.cxx vtkTest.cxx
>  )
>
> AlgorithmSourceExample.xml:
>
> 
>   
> label="AlgorithmSourceExample">
>   long_help="AlgorithmSourceExample"
> short_help="AlgorithmSourceExample">
>  
>
>  
> 
>
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Search the list archives at: http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview

[Paraview] plugin error: "Input for algorithm vtkPVPostFilter is of type ..., but a vtkPolyData is required"

[corinna reuter]

To anybody experienced in plugin programming:

I am trying to include items in the paraview tree, which are not derived
from vtkDataSet or vtkDataObject. For example, a mySQL database should be
allowed as a data source, which delivers data in an intermediate format.
Next I'd implement filters which take the intermediate format and produce
vtkDataSets as output.

Is it possible to use such "abstract" (not vtkDataSet) objects in the
paraview tree as a representation of the intermediate format? I have tried
to integrate the example for custom classes from
http://www.vtk.org/Wiki/VTK_Examples_vtkAlgorithm_Source into a paraview
plugin (CMakeLists and servermanager xml below). When the plugin is loaded
and I add a source of the new type AlgorithmSourceExample, this error is
thrown:

ERROR: In
/opt/ParaView-v5.0.1/VTK/Common/ExecutionModel/vtkDemandDrivenPipeline.cxx,
line 810
vtkPVPostFilterExecutive (0x3857810): Input for connection index 0 on input
port index 0 for algorithm vtkPVPostFilter(0x3858910) is of type vtkTest,
but a vtkPolyData is required.

What do I need to change such that the vtkTest objects, produced by
AlgorithmSourceExample, can live in the paraview tree?

Thank you
Corinna

CMakeLists.txt:

FIND_PACKAGE(ParaView REQUIRED)
INCLUDE(${PARAVIEW_USE_FILE})
#changes required in
http://www.vtk.org/Wiki/VTK_Examples_vtkAlgorithm_Source:
# remove vtkCxxRevisionMacro calls
# change vtkTypeRevisionMacro into vtkTypeMacro
# remove call output->SetPipelineInformation( outInfo );
 ADD_PARAVIEW_PLUGIN(
   AlgorithmSourceExample
  "1.0"
   SERVER_MANAGER_XML AlgorithmSourceExample.xml
   SERVER_MANAGER_SOURCES vtkTestSource.cxx vtkTest.cxx
 )

AlgorithmSourceExample.xml:


  
   
 
 
   
 

___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Search the list archives at: http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview

Re: [Paraview] Automatically updating time variable dataset

[corinna reuter]

This is not practical. We have tried this before. The data are updated in a
background thread, so that data to be displayed are always available
without any delay. The continuous redrawing uses 100% cpu. Even with a
refresh rate reduced to 20 updates per second (sleep commands) cpu usage is
still 70%. Slower refresh rates cause unwanted effects for calculations
that vary quickly. This load is predominantly caused by the communication
overhead and almost independent of the actual rendering work (tested with a
50x50 pixel window and single triangle dataset).

Does anybody have a real solution how a background thread can wake up the
main thread when data has changed? A continuous trigger from the GUI wastes
too many resources. An event based mechanism is needed.

Conny


On Wed, Jun 22, 2016 at 12:59 PM, Mathieu Westphal <
mathieu.westp...@kitware.com> wrote:

> Hello
>
> A work around would be to declare two (fake) timestep in your reader and
> play the animation in loop in paraview.
>
> Regards,
>
> Mathieu Westphal
>
> On Wed, Jun 22, 2016 at 12:51 PM, corinna reuter <
> corinnareute...@gmail.com> wrote:
>
>> A question to the specialists:
>> Can Paraview refresh its view automatically when a dataset updates its
>> state?
>>
>> A plugin starts a background thread that calculates new dataset values
>> periodically. Each time a calculation step is finished, all filters and
>> views connected to the generated dataset should be updated. (Something
>> similar is done in a live catalyst connection. But catalyst has a serious
>> drawback for large datasets. It runs in a separate process, so that the
>> datasets cannot be in shared memory, but have to be transferred, requiring
>> redundant storage space and wasted network bandwidth.) Although paraview is
>> said to be not thread safe, it should't be problematic as long as access to
>> VTK classes is limited to the main thread. The difficult question is, how
>> can paraview be notified that a plugin has updated its output while the
>> main thread only waits for GUI or network events when idle?
>>
>> Any ideas or recommendations?
>> Conny
>>
>> ___
>> Powered by www.kitware.com
>>
>> Visit other Kitware open-source projects at
>> http://www.kitware.com/opensource/opensource.html
>>
>> Please keep messages on-topic and check the ParaView Wiki at:
>> http://paraview.org/Wiki/ParaView
>>
>> Search the list archives at: http://markmail.org/search/?q=ParaView
>>
>> Follow this link to subscribe/unsubscribe:
>> http://public.kitware.com/mailman/listinfo/paraview
>>
>>
>
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Search the list archives at: http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview

[Paraview] Automatically updating time variable dataset

[corinna reuter]

A question to the specialists:
Can Paraview refresh its view automatically when a dataset updates its
state?

A plugin starts a background thread that calculates new dataset values
periodically. Each time a calculation step is finished, all filters and
views connected to the generated dataset should be updated. (Something
similar is done in a live catalyst connection. But catalyst has a serious
drawback for large datasets. It runs in a separate process, so that the
datasets cannot be in shared memory, but have to be transferred, requiring
redundant storage space and wasted network bandwidth.) Although paraview is
said to be not thread safe, it should't be problematic as long as access to
VTK classes is limited to the main thread. The difficult question is, how
can paraview be notified that a plugin has updated its output while the
main thread only waits for GUI or network events when idle?

Any ideas or recommendations?
Conny
___
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Search the list archives at: http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview