Difference between revisions of "Engineering:TCON 2009"

From NAMIC Wiki
Jump to: navigation, search
Line 2: Line 2:
  
 
==02-12-2009==
 
==02-12-2009==
*Topic #1: NA-MIC data repository
+
*Agenda:
*Topic #2: Python wrapping of ITK in Slicer: adding vtk-itk glue pipeline connection filters to Slicer build, so Python scripts can use ITK filters.
+
**Topic #1: NA-MIC data repository
 +
**Topic #2: Python wrapping of ITK in Slicer: adding vtk-itk glue pipeline connection filters to Slicer build, so Python scripts can use ITK filters.
 +
* Attendees: Jim Miller, Tina Kapur, Nicole Aucoin, Kurt Lisle, Ron Kikinis, Clement, Dan Marcus, Katie Hayes, Wendy Plesniak, Xiodong, Vincent Magnotta, Steve Pieper, Will Schroeder
 +
 
 +
* NAMIC data rep:
 +
* there is namic data, test data and tutorial data floating around in various places (including in Slicer download, different data bases), Ron suggests that we think about how we should organise it into one place for nightly builds, tutorials etc.
 +
* Wendy: we have an instance of XND web services running on xnd.slicer.org, we could use it
 +
** Steve: it's on an Amazon cloud, fine for testing, but not good for a high availability server. Should be where BWH sysadmins can monitor it. Shouldn't be hidden by a firewall. Need something hardened.
 +
** also using central.xnat.org, Midas
 +
** Dan: says right long term place for it is XNAT central, but right now Slicer is programmed against the desk top API. He suggests putting everything in a desktop and then porting it to central. WJP: could put .xcat files on central, Slicer3 can read them automatically.
 +
** Ron: manual upload is okay for now, need reliable downloads
 +
** Steve: for some tutorial and sample data put links in the GUI, ie file menu, sample data. Dan: start with XNAT Central, then.
 +
** Ron: Two areas: tutorial data sets, compiled test data sets. For read, completely open.
 +
** Use cases: Slicer 101 training page, link to data could point directly into xnat central. Nightly build: been using SVN, but the data is being packaged up for distribution right now, would like to avoid depending on another server, make it clear as to if need to install testing data (assume for now stay in source tree, but do need to test installed version - dl data in that case).
 +
** Clean up: take out extra data files, .h files, if not needing to build extra stuff against an installed Slicer (has been abandoned due to different compilers?).  Need guidelines for the developers as to what should be installed.
 +
*** If someone wants to write a CLI or loadable module, they have to have a built version of Slicer to do that
 +
** Action items: Create a Slicer tutorial data project on xnat central and upload to it, make it open. Make xnat central so can access open projects w/o having to log in at all. Can it be done within 2 days? Dan says: Monday. He'll send an email when it's set up, make a wiki page on how to upload it.
 +
** For the test data sets, keep using SVN, but encourage developers to make the smallest possible test data set
 +
*** Add an upper limit for test data size, over it it should go in central.xnat.org
 +
*** Add documentation for the test data sets somewhere. In directory as a ReadMe.txt file?
 +
*** Make sure fill in doxygen style comments for auto generated documentation
 +
** There will be a new Slicer 3.4 tutorial page, everything will be on the central xnat repository
 +
 
 +
* Topic 2 : Curt: wrap ITK in python.
 +
** To be able to move Slicers python console and ITK filters en mass for experimental prototyping.
 +
** Historic vtkITK project, hand wrappers for indiv filters want to us, and CLI mechanism so can put in ITK filters directly.
 +
** If we do auto wrapping, the wrapping is there, but there's no vtkITK pipeline connection available.
 +
** Steve: You can go into ITK build and turn on the wrapping, Vince says builds take really long on Windows, so hesitant to turn it on by default in Slicer build.
 +
** in python console can instantiate an ITK filter, but getting things into and out of it is a problem. Curt envisions loading a data set into Slicer and trying out different filters to see what you want to put in a CLI. Can try using the numpy? get a numpy array of slicer image data and manipulate it w/numpy or use as a vtk array? Something else needed? Checking Luca's old emails... the Slicer way of wrapping for python is different than the native VTK way of wrapping python. Luca would like some examples from Curt, use cases. There is a way to go from Slicer mrml to numpy array to ITK. Jim could embed this into a helper.
 +
** Jim: want to grab a mrml node and say get the ITK representation of the data in that node (have the orientations come across, and meta data in mrml node that ITK can represent, LPS/RAS conversion).
 +
** no action items just yet, Curt will explore and report what is needed to make wrappers
 +
** Luca: need to be clear when working with VTK wrapped things versus Slicer wrapped things, can't go from one to the other, but each one individually should be able to do .
 +
** Will: some email on the ParaView list about wrapping, he'll send stuff to Curt and Steve.
 +
** building python problems on Dash18: connection problems with the scipy svn server. Mirror?
 +
 
 +
* Dave Cole has been working with Gordon, has teem building with CMake, working on a "proper" Mac release (new feature?)
 +
* Jim updating a couple of builds today, errors building from getbuildtest.tcl, source tree is checked out of someplace else (teem, igt link), need to remove the directory and rebuild.
 +
* Steve: latest windows build installer can bring up the python console. Katie reports seg faults on Darwin (unrelated?).
 +
** there are python versions 2.6 and 3.0, but consensus here has been to stick with 2.5 for numpy and other packages.
 +
* Been going through the bug tracker and clean up things for the release of Slicer 3.4
 +
* Was there a recent workshop that Randy Gollub was involved in Puerto Rico? mBIRN spring last year
  
 
==02-05-2009==
 
==02-05-2009==

Revision as of 20:54, 12 February 2009

Home < Engineering:TCON 2009

Back to TCON:Main, Events

02-12-2009

  • Agenda:
    • Topic #1: NA-MIC data repository
    • Topic #2: Python wrapping of ITK in Slicer: adding vtk-itk glue pipeline connection filters to Slicer build, so Python scripts can use ITK filters.
  • Attendees: Jim Miller, Tina Kapur, Nicole Aucoin, Kurt Lisle, Ron Kikinis, Clement, Dan Marcus, Katie Hayes, Wendy Plesniak, Xiodong, Vincent Magnotta, Steve Pieper, Will Schroeder
  • NAMIC data rep:
  • there is namic data, test data and tutorial data floating around in various places (including in Slicer download, different data bases), Ron suggests that we think about how we should organise it into one place for nightly builds, tutorials etc.
  • Wendy: we have an instance of XND web services running on xnd.slicer.org, we could use it
    • Steve: it's on an Amazon cloud, fine for testing, but not good for a high availability server. Should be where BWH sysadmins can monitor it. Shouldn't be hidden by a firewall. Need something hardened.
    • also using central.xnat.org, Midas
    • Dan: says right long term place for it is XNAT central, but right now Slicer is programmed against the desk top API. He suggests putting everything in a desktop and then porting it to central. WJP: could put .xcat files on central, Slicer3 can read them automatically.
    • Ron: manual upload is okay for now, need reliable downloads
    • Steve: for some tutorial and sample data put links in the GUI, ie file menu, sample data. Dan: start with XNAT Central, then.
    • Ron: Two areas: tutorial data sets, compiled test data sets. For read, completely open.
    • Use cases: Slicer 101 training page, link to data could point directly into xnat central. Nightly build: been using SVN, but the data is being packaged up for distribution right now, would like to avoid depending on another server, make it clear as to if need to install testing data (assume for now stay in source tree, but do need to test installed version - dl data in that case).
    • Clean up: take out extra data files, .h files, if not needing to build extra stuff against an installed Slicer (has been abandoned due to different compilers?). Need guidelines for the developers as to what should be installed.
      • If someone wants to write a CLI or loadable module, they have to have a built version of Slicer to do that
    • Action items: Create a Slicer tutorial data project on xnat central and upload to it, make it open. Make xnat central so can access open projects w/o having to log in at all. Can it be done within 2 days? Dan says: Monday. He'll send an email when it's set up, make a wiki page on how to upload it.
    • For the test data sets, keep using SVN, but encourage developers to make the smallest possible test data set
      • Add an upper limit for test data size, over it it should go in central.xnat.org
      • Add documentation for the test data sets somewhere. In directory as a ReadMe.txt file?
      • Make sure fill in doxygen style comments for auto generated documentation
    • There will be a new Slicer 3.4 tutorial page, everything will be on the central xnat repository
  • Topic 2 : Curt: wrap ITK in python.
    • To be able to move Slicers python console and ITK filters en mass for experimental prototyping.
    • Historic vtkITK project, hand wrappers for indiv filters want to us, and CLI mechanism so can put in ITK filters directly.
    • If we do auto wrapping, the wrapping is there, but there's no vtkITK pipeline connection available.
    • Steve: You can go into ITK build and turn on the wrapping, Vince says builds take really long on Windows, so hesitant to turn it on by default in Slicer build.
    • in python console can instantiate an ITK filter, but getting things into and out of it is a problem. Curt envisions loading a data set into Slicer and trying out different filters to see what you want to put in a CLI. Can try using the numpy? get a numpy array of slicer image data and manipulate it w/numpy or use as a vtk array? Something else needed? Checking Luca's old emails... the Slicer way of wrapping for python is different than the native VTK way of wrapping python. Luca would like some examples from Curt, use cases. There is a way to go from Slicer mrml to numpy array to ITK. Jim could embed this into a helper.
    • Jim: want to grab a mrml node and say get the ITK representation of the data in that node (have the orientations come across, and meta data in mrml node that ITK can represent, LPS/RAS conversion).
    • no action items just yet, Curt will explore and report what is needed to make wrappers
    • Luca: need to be clear when working with VTK wrapped things versus Slicer wrapped things, can't go from one to the other, but each one individually should be able to do .
    • Will: some email on the ParaView list about wrapping, he'll send stuff to Curt and Steve.
    • building python problems on Dash18: connection problems with the scipy svn server. Mirror?
  • Dave Cole has been working with Gordon, has teem building with CMake, working on a "proper" Mac release (new feature?)
  • Jim updating a couple of builds today, errors building from getbuildtest.tcl, source tree is checked out of someplace else (teem, igt link), need to remove the directory and rebuild.
  • Steve: latest windows build installer can bring up the python console. Katie reports seg faults on Darwin (unrelated?).
    • there are python versions 2.6 and 3.0, but consensus here has been to stick with 2.5 for numpy and other packages.
  • Been going through the bug tracker and clean up things for the release of Slicer 3.4
  • Was there a recent workshop that Randy Gollub was involved in Puerto Rico? mBIRN spring last year

02-05-2009

Agenda:

  • Attendees: Jim, Wendy, Tina, Ron, Nicole, Steve, Curt, Vince

Feature freeze occurred on 2/4/09 - we should be only cleaning up, fixing bugs, and writing documentation for existing Slicer modules.

Target for code freeze will be at the end of February. Please get bug fixes in during February, if possible. We will start making release candidate builds on or around March 1st.

Everyone please check the module documentation pages and make sure to fill out the documentation. Ron has added candidate names for individuals responsible for the modules. Please use the template and go along with the naming scheme. Some reorganization of the list may occur. Email will go out for everyone to finish their module in the next week or so.

Compile a list of grants that have benefitted from NA-MIC. Both active and expired grants which have benefitted should be entered. Collaborations with other groups is good. Joint publications are preferred to demonstrate effective collaboration. We are also trying to show that NA-MIC people are involved in other productive activities, as well.

Core 2 NAC and BIRN activity should be listed.

Collaboration grants should be added to be listed. We are looking to list solid grants or efforts that should be listed. Send to Tina or add to this list.

  • New Tools

New tools for registration and atlas development have been recently released. UCLA NCBC effort or related grant may have assisted with this (James Gee). Jim Miller evaluating to decide if these could make useful Slicer plug-ins. They are currently command line tools which could be adapted for use with Slicer. Is this a benefit to show that this NCBC's tool products are compatible with Slicer.

  • Python building in Windows

Steve is experiencing some Python build problems under Windows related to GenerateCLP. Jim and Steve discussed how to resolve.

iPython is now available for use with Slicer. There is a small installation script to run after Slicer is built. WrapITK is the newest way to wrap ITK in Python. vtkITK is the older way. All of ITK can be wrapped automatically, but this option is turned off by default, since it slows down the build.

01-29-2009

Attendees: Jim, Steve, Tina, Dan, Nicole

01-22-2009

Attendees: Jim, Steve, Tina, Wendy, Xiadong, Nicole

  • collaboration grants - doesn't look like anything is going in feb

01-15-2009

Attendees: Jim, Steve, Katie, Tina, Wendy, Nicole, Vince, Xiaodong, Sebastien

Agenda:

  • LGPL QT - should we consider this going forward?
  • Slicer code freeze on Feb 4th. Slicer3 Events for requirements for modules to be included.