[vtkusers] python exceptions patch needed for vtk 5.6
Florian Bruckner
e0425375 at gmail.com
Thu Jun 24 05:21:22 EDT 2010
Thanks! your code partly solve my problem but i have still one problem:
the code with the observer seems to work with vtkAVSucdReader as well as
vtkUnstructuredGridReader.
But vtkXMLUnstructuredGridReader seems to produce 3 error events and
actually only the 2. is caught by the observer!? if I use an empty file
as input I get the following 3 messages without observer:
ERROR: In /home/florian/magnet/tools/vtk-5.6/VTK/IO/vtkXMLParser.cxx,
line 484
vtkXMLDataParser (0xee0170): Error parsing XML in stream at line 1,
column 0, byte index -1: no element found
ERROR: In /home/florian/magnet/tools/vtk-5.6/VTK/IO/vtkXMLReader.cxx,
line 363
vtkXMLUnstructuredGridReader (0xec8ab0): Error parsing input file.
ReadXMLInformation aborting.
ERROR:
In /home/florian/magnet/tools/vtk-5.6/VTK/Filtering/vtkExecutive.cxx,
line 757
vtkStreamingDemandDrivenPipeline (0xed41b0): Algorithm
vtkXMLUnstructuredGridReader(0xec8ab0) returned failure for request:
vtkInformation (0xedd630)
Debug: Off
Modified Time: 70
Reference Count: 1
Registered Events: (none)
Request: REQUEST_INFORMATION
FORWARD_DIRECTION: 0
ALGORITHM_AFTER_FORWARD: 1
the observer only catches the "Error parsing input file" error. The
other messages are directly printed to stdout.
is this a bug? or is something missing?
thanks
FloB
On Tue, 2010-06-01 at 11:36 -0400, Lic. José M. Rodriguez Bacallao
wrote:
> thanks, I will analyze that later, thanks for your answer!!!
>
> On Tue, Jun 1, 2010 at 10:30 AM, David Gobbi <david.gobbi at gmail.com> wrote:
> > The CallDataType attribute is read by the vtkPythonCommand, which is
> > working behind the scenes. You can read vtkPythonCommand.cxx to see
> > how it works. I think that a having a method called GetCallDataType()
> > would be easier for people to understand than having an attribute.
> >
> > David
> >
> >
> > On Tue, Jun 1, 2010 at 7:11 AM, Lic. José M. Rodriguez Bacallao
> > <jmrbcu at gmail.com> wrote:
> >> well, I can confirm that Charl's patch work very well. The only thing
> >> to worry is from raising an exception within an observer in python but
> >> for me this is not a major problem. Right now, I am using this patch
> >> with vtk 5.4.2 because is nice to see python exceptions and for my
> >> threading issues (the last is a big issue for me), this is the only
> >> patch I have seen for vtk python bindings that address the GIL issue.
> >> I have tested this patch in linux and soon will be doing the same with
> >> windows because my application must support windows/linux and possible
> >> MacOSX.
> >>
> >> David, I like your way of catching errors but ins this code:
> >>
> >> ==== python code ====
> >> import vtk
> >>
> >> class ErrorObserver:
> >>
> >> def __init__(self):
> >> self.__ErrorOccurred = False
> >> self.__ErrorMessage = None
> >> self.CallDataType = 'string0'
> >>
> >> def __call__(self, obj, event, message):
> >> self.__ErrorOccurred = True
> >> self.__ErrorMessage = message
> >>
> >> def ErrorOccurred(self):
> >> occ = self.__ErrorOccurred
> >> self.__ErrorOccurred = False
> >> return occ
> >>
> >> def ErrorMessage(self):
> >> return self.__ErrorMessage
> >>
> >> e = ErrorObserver()
> >>
> >> a = vtk.vtkImageReader()
> >> a.AddObserver('ErrorEvent', e)
> >>
> >> print "doing update"
> >> a.Update()
> >> if e.ErrorOccurred():
> >> print e.ErrorMessage()
> >> ==== python code ====
> >>
> >> where do you use self.CallDataType?
> >>
> >>
> >> On Sun, May 30, 2010 at 9:03 PM, David Gobbi <david.gobbi at gmail.com> wrote:
> >>> Hi Florian,
> >>>
> >>> Here is an example of how to make an error observer in python. Most
> >>> of the code is straightforward. The __call__ method is needed because
> >>> VTK-Python observers must be callable objects. The only messy part is
> >>> the "CallDataType", which is needed because in C++ the VTK object
> >>> passes the error message as a "void *" instead of as a properly typed
> >>> object, so the python observer needs a hint that the "void *" is a
> >>> "char *".
> >>>
> >>> ==== python code ====
> >>> import vtk
> >>>
> >>> class ErrorObserver:
> >>>
> >>> def __init__(self):
> >>> self.__ErrorOccurred = False
> >>> self.__ErrorMessage = None
> >>> self.CallDataType = 'string0'
> >>>
> >>> def __call__(self, obj, event, message):
> >>> self.__ErrorOccurred = True
> >>> self.__ErrorMessage = message
> >>>
> >>> def ErrorOccurred(self):
> >>> occ = self.__ErrorOccurred
> >>> self.__ErrorOccurred = False
> >>> return occ
> >>>
> >>> def ErrorMessage(self):
> >>> return self.__ErrorMessage
> >>>
> >>> e = ErrorObserver()
> >>>
> >>> a = vtk.vtkImageReader()
> >>> a.AddObserver('ErrorEvent', e)
> >>>
> >>> print "doing update"
> >>> a.Update()
> >>> if e.ErrorOccurred():
> >>> print e.ErrorMessage()
> >>> ==== python code ====
> >>>
> >>>
> >>>
> >>> On Sun, May 30, 2010 at 3:47 AM, Florian Bruckner <e0425375 at gmail.com> wrote:
> >>>> OK, for me this means that I will use vtk observers, since I need a portable
> >>>> solution that runs on a default vtk installation.
> >>>>
> >>>> @gobbi: could you please send me a short example, of how you use these
> >>>> observers in python?
> >>>>
> >>>> nevertheless it would be nice to see this patch in some future vtk releases.
> >>>> If the only problem are these unreachable code warnings in windows, why
> >>>> can't you simply add some helper-functions that execute this otherwise
> >>>> unreachable code? Since I only use Linux I will not be able to test it for
> >>>> myself, but I can't imagine that none of the windows programmers could find
> >>>> a solution for this. perhaps you should give it another try and repost the
> >>>> patch to the developers list?
> >>>>
> >>>> thanks to all for your help
> >>>> I wish you what
> >>>> FloB
> >>>>
> >>>>
> >>>> On Sat, May 29, 2010 at 4:36 AM, David Gobbi <david.gobbi at gmail.com> wrote:
> >>>>>
> >>>>> Hi Charl,
> >>>>>
> >>>>> I've looked over the patch and I have one question. What happens when
> >>>>> an python exception is thrown in a python observer callback? I didn't
> >>>>> see any modifications to vtkPythonCommand, so I'm guessing that the
> >>>>> error will just be printed (like it is now) and not fed back to the
> >>>>> original caller.
> >>>>>
> >>>>> David
> >>>>>
> >>>>>
> >>>>> On Fri, May 28, 2010 at 4:43 PM, Charl Botha <c.p.botha at tudelft.nl> wrote:
> >>>>> > Dear Florian and David,
> >>>>> >
> >>>>> > On 28 May 2010 21:07, Florian Bruckner <e0425375 at gmail.com> wrote:
> >>>>> >>
> >>>>> >> is there a reason why the patch was not merged into the official vtk
> >>>>> >> release?
> >>>>> >>
> >>>>> >> i'm just wondering because that patch is fairly old and for me this
> >>>>> >> seems to be a very good solution for a general problem.
> >>>>> >
> >>>>> > The patch has a permanent home here:
> >>>>> >
> >>>>> > http://code.google.com/p/devide/source/browse/trunk/johannes/patches/pyvtk_tryexcept_and_pyexceptions.diff
> >>>>> >
> >>>>> > Of course I think it's a good patch. :) It does the following:
> >>>>> > 1. Turns all VTK errors into Python exceptions.
> >>>>> > 2. Carefully releases the GIL during all calls into VTK (grabbing it
> >>>>> > back for observers) so that you can successfully explicitly
> >>>>> > multi-thread VTK pipelines from Python.
> >>>>> > 3. Tries to trap bad_alloc.
> >>>>> >
> >>>>> > I've been working on this for years, but due to Windows unreachable
> >>>>> > code warnings that I have not been able to solve, I've never committed
> >>>>> > this to VTK (it works, just gives many warnings during compilation).
> >>>>> > My own binaries all have this patch integrated (DeVIDE [1] and VTK
> >>>>> > [2]).
> >>>>> >
> >>>>> > The current version is for 5.4.2, I'd like to update for 5.6 as soon
> >>>>> > as I can figure out how to check out a clean 5.6 with git (until very
> >>>>> > recently there was no tag yet, I see that it's there now!
> >>>>> > http://vtk.org/gitweb?p=VTK.git;a=summary )
> >>>>> >
> >>>>> > I'm also considering to make a wiki page on devide.googlecode.com that
> >>>>> > documents the patch.
> >>>>> >
> >>>>> > See you,
> >>>>> > Charl
> >>>>> >
> >>>>> > [1] http://graphics.tudelft.nl/Projects/DeVIDE
> >>>>> > [2] http://cpbotha.net/software/latest-vtk-windows-binaries/
> >>>>> >
> >>>>> >
> >>>>> >>
> >>>>> >> On Fri, May 28, 2010 at 4:07 PM, David Gobbi <david.gobbi at gmail.com>
> >>>>> >> wrote:
> >>>>> >>>
> >>>>> >>> Hi Florian,
> >>>>> >>>
> >>>>> >>> I haven't seen mention of the patch here on the list, and I know
> >>>>> >>> that it isn't in VTK 5.6 or in VTK devel. For my own apps, I use
> >>>>> >>> VTK observers to catch reader errors.
> >>>>> >>>
> >>>>> >>> David
> >>>>> >>>
> >>>>> >>>
> >>>>> >>> On Fri, May 28, 2010 at 12:52 AM, Florian Bruckner
> >>>>> >>> <e0425375 at gmail.com> wrote:
> >>>>> >>> > hi,
> >>>>> >>> >
> >>>>> >>> > I just figured out the problem that no python exceptions are raised
> >>>>> >>> > if
> >>>>> >>> > my reader produces an error. I use vtk-5.6.
> >>>>> >>> >
> >>>>> >>> > after searching the net and the mailing list I found that there are
> >>>>> >>> > some
> >>>>> >>> > patches available
> >>>>> >>> > (turn-vtk-errors-into-python-exceptions-and-catch-bad-malloc patch)
> >>>>> >>> > which should solve this problem. Unfortunately I could neither find
> >>>>> >>> > such
> >>>>> >>> > a patch for vtk-5.6 nor any actual posts about this topic.
> >>>>> >>> >
> >>>>> >>> > So should this patch allready be included in the new vtk-5.6? or
> >>>>> >>> > where
> >>>>> >>> > can I get it otherwise? are there any reasons why the patch is not
> >>>>> >>> > included in the official release? or is there another work around
> >>>>> >>> > that
> >>>>> >>> > is proposed for this kind of problems?
> >>>>> >>> >
> >>>>> >>> > thanks
> >>>>> >>> > FloB
> >>>>> >>
> >>>>> >>
> >>>>> >> _______________________________________________
> >>>>> >> 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 VTK FAQ at:
> >>>>> >> http://www.vtk.org/Wiki/VTK_FAQ
> >>>>> >>
> >>>>> >> Follow this link to subscribe/unsubscribe:
> >>>>> >> http://www.vtk.org/mailman/listinfo/vtkusers
> >>>>> >>
> >>>>> >
> >>>>> >
> >>>>> >
> >>>>> > --
> >>>>> > dr. charl p. botha :: http://graphics.tudelft.nl/CharlBotha
> >>>>> > work: cpbotha at medvis.org personal: cpbotha at cpbotha.net
> >>>>> >
> >>>>> > See our latest papers on shape spaces, brain connectivity, shoulder
> >>>>> > segmentation, human motion analysis, voxel classification and eye fat
> >>>>> > deformation! http://graphics.tudelft.nl/Publications/2010
> >>>>> >
> >>>>
> >>>>
> >>> _______________________________________________
> >>> 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 VTK FAQ at: http://www.vtk.org/Wiki/VTK_FAQ
> >>>
> >>> Follow this link to subscribe/unsubscribe:
> >>> http://www.vtk.org/mailman/listinfo/vtkusers
> >>>
> >>
> >>
> >>
> >> --
> >> Lic. José M. Rodriguez Bacallao
> >> Centro de Biofisica Medica
> >> -----------------------------------------------------------------
> >> Todos somos muy ignorantes, lo que ocurre es que no todos ignoramos lo mismo.
> >>
> >> Recuerda: El arca de Noe fue construida por aficionados, el titanic
> >> por profesionales
> >> -----------------------------------------------------------------
> >>
> >
>
>
>
More information about the vtkusers
mailing list