[vtk-developers] DoublePi proposal

Marcus D. Hanwell marcus.hanwell at kitware.com
Tue May 15 19:55:33 EDT 2012


Andrew, Bill,

We have been working on getting ParaView building against VTK modular,
and fixing up issues as they cropped up. We have also spent some time
fixing up the tests that are running and adding a few in. I also had
other projects that had been neglected I have been getting back to.

I plan on spending some time on restoring functionality/adding more
tests Wednesday/Thursday. I would certainly welcome input (and
patches) from those that know the code in these modules best. Failing
that Chris and I will be working on restoring more tests now that
ParaView is mostly ported over (master as of Friday is building
against modular VTK 6).

We would also like to work on getting the less reliable tests weeded
out so that we can have a green dashboard. Ideally I would like to
compile a list of people responsible/knowledgeable on the modules that
have been established. We were thinking of adding this list to a wiki
page, this would help both with code review and quality control.

Just a few thoughts/input on what we have been up to.

Thanks,

Marcus

On Tue, May 15, 2012 at 7:20 PM, Andrew Maclean
<andrew.amaclean at gmail.com> wrote:
> Hi Bill,
> I had noticed this. I have been assuming that Marcus or someone has been
> gradually reintroducing the tests.
> If this is not the case is there something I should do?
>
> Andrew
>
>
> On Wed, May 16, 2012 at 9:12 AM, Bill Lorensen <bill.lorensen at gmail.com>
> wrote:
>>
>> Andrew,
>>
>> Looks to me like the Parametric classes are not being tested by VTK6
>>
>> Bill
>>
>> On Tue, May 15, 2012 at 6:47 PM, Andrew Maclean
>> <andrew.amaclean at gmail.com> wrote:
>> > Hi Bill, David,
>> >   As one of you pointed out earlier, the VTK Parametric classes will be
>> > affected by this change. In particular, the regression tests will
>> > fail due to the higher precision of Pi. I would only expect that the
>> > pictures for the regression tests would need to be updated.
>> >
>> > When I wrote these in conjunction with Will, I must admit I was
>> > frustrated
>> > that vtkMath::Pi only returned float. However I made a decision to use
>> > vtkMath::Pi rather than yet another constant for Pi in the hopes that
>> > this
>> > would change in the future. The greater precision should improve the
>> > look of
>> > some of the surfaces. E.g. the little imperfection at the pole of the
>> > ellipsoid should get smaller.
>> >
>> > Let me know if I can help in any way.
>> >
>> > When things settle, I will also update the PDF titled "Parametric
>> > Equations
>> > for Surfaces" in the documentation to VTK 6.
>> >
>> > Regards
>> >    Andrew
>> >
>> >
>> >
>> >>
>> >>
>> >> ---------- Forwarded message ----------
>> >> From: David Gobbi <david.gobbi at gmail.com>
>> >> To: Bill Lorensen <bill.lorensen at gmail.com>
>> >> Cc: VTK Developers <vtk-developers at vtk.org>
>> >> Date: Mon, 14 May 2012 16:38:37 -0600
>> >> Subject: Re: [vtk-developers] DoublePi proposal
>> >> Bill, the main reason that I put my patch on hold was that I wanted to
>> >> wait until VTK 6 had reached at least the same level of test coverage
>> >> as VTK 5.  I still think that this is a valid reason for keeping this
>> >> patch on hold.  The added pi precision will cause subtle changes in
>> >> many regression images, and some of the changes will be significant
>> >> enough to cause test failures.  That's great if all the regression
>> >> tests are running, because we can examine any test failures that occur
>> >> and replace the regression images as necessary.
>> >>
>> >> But right now in VTK 6, about half of our regression tests aren't
>> >> running.  So if this Pi patch is pushed now, we won't see all the
>> >> regression test failures immediately.  We'll see some regression test
>> >> failures some weeks (or months?) later as the various tests are
>> >> re-activated, and we won't be sure if the failures were caused by this
>> >> patch, or if they were caused by some other patch.
>> >>
>> >>  - David
>> >>
>> >>
>> >> On Mon, May 14, 2012 at 11:59 AM, David Gobbi <david.gobbi at gmail.com>
>> >> wrote:
>> >> > http://review.source.kitware.com/#/t/427/
>> >> >
>> >> >
>> >> > On Mon, May 14, 2012 at 10:50 AM, Bill Lorensen
>> >> > <bill.lorensen at gmail.com> wrote:
>> >> >> David,
>> >> >>
>> >> >> Can you point me to the gerrit topic? If not, I can start from
>> >> >> scratch.
>> >> >>
>> >> >> Bill
>> >
>> > --
>> > ___________________________________________
>> > Andrew J. P. Maclean
>> >
>> > ___________________________________________
>>
>>
>>
>> --
>> Unpaid intern in BillsBasement at noware dot com
>
>
>
>
> --
> ___________________________________________
> Andrew J. P. Maclean
>
> ___________________________________________



More information about the vtk-developers mailing list