[Paraview] stability of paraview

[Will Schroeder]

Let me add a couple of factoids to what Andy has stated about the 
software process:
+ this process is part of an overall process which we are using to 
develop FDA-approved software.
+ CMake (and parts of the process) have been adopted by the KDE 
community, which is one of, if not the largest, open source projects in 
the world (KDE is the GUI desktop environment for Linux)
+ Andy refers to NAMIC (na-mic.org) which one of seven US National 
Centers for Biomedical Imaging (funded by National Institutes of 
Health). These 5-10 year long centers are funded at a level of 
approximate $4/million year (depending on the particular center). 
Kitware plays a central role in the NAMIC NCBC principally because of 
the software process.
+ ParaView in some respects is relatively young. As Berk has indicated, 
we are moving to a Qt-based GUI shortly, in large part due to our 
friends at the Nat'l labs (Sandia, Livermore, Los Alamos) who are 
partners in the development of ParaView. And as you can see from the 
mailing list, we have some passionate community members---this is 
important because the software process works best when we can focus the 
feedback and contributions from an active community.

Will


Andy Cedilnik wrote:
> Hi Peter,
>
> I will let other people talk about the stability of ParaView, but I 
> will talk about the software process behind ParaView and other Kitware 
> products.
>
> I guess the most important aspect of any products is that what you 
> test works perfectly. In ParaView's case, we run about 300 tests every 
> night that test ParaView specific operations on several different 
> platforms. This is combined with about 800 tests of the Visualization 
> Toolkit VTK, which test the VTK's capabilities.
>
> These tests are run every night on all systems and the results are 
> submitted to the web page called Testing Dashboard:
>
> http://public.kitware.com/dashboard.php?name=paraview
>
> This way anybody (developer or user) can look at the current state of 
> the project.
>
> If you look at it today, you can see some issues that are mostly 
> caused by our current effort towards the upcoming ParaView 2.6 and the 
> next major release 3.0.
>
> Beyond ParaView, Kitware is a leader in several other open-source 
> community driven projects. In addition to this, Kitware plays a vital
> role in the National Alliance for Medical Image Computing, by leading 
> the engineering and infrastructure efforts.
>
> Large projects like these cannot be developed and maintained without a 
> sophisticated software process. We use a novel process based on 
> methods from agile development and test-driven development. The 
> process incorporates source code control (using tools such as CVS or 
> Subversion), build management using CMake, and regression testing 
> using CTest and DART. One of the highlights of this process is that it 
> runs continuously, simultaneously on multiple platforms, and produces 
> its results on web-accessible dashboards. Thus as software is checked 
> into the source code repository, it is immediately tested and results 
> posted. Developers rapidly obtain feedback on their changes and 
> correct problems immediately.
>
> Furthermore the software development process is augmented with 
> Internet based collaboration tools. Mailing lists are the most used 
> source of information for majority of collaborators. They offer large 
> pool of expertise, as well as the source of frequently asked 
> questions. They are archived and can be searched. Secondly, there are 
> Wiki Web pages, which are community editable Web pages. Each page has 
> an Edit button, where anybody can add or update content. Along with 
> this, we rely heavily on a set of static and automatically generated 
> web pages. These web pages include the regular information about the 
> project, such as documentation, where to download it, and other 
> resources. Along with that, there are parts generated periodically, 
> such as application program interface (API) documentation.
>
> Let me know if you have any question.
>
> Thank you.
>
>       Andy
>
> Peter J. Bismuti wrote:
>> Hi, here at Boeing we're contemplating using Paraview as one tool for 
>> visualizing solutions from an in-house CFD code (Navier-Stokes). 
>> There is concern that an open-source produce won't have an acceptable 
>> level of QA.  There have been some reports of frequent crashes.  
>> Anybody care to share their experiences or opinions? 
>> Thanks
>> Peter Bismuti
>> Boeing
>>   
>