[vtk-developers] vtkImageConvolve -- why is it there?

[Will Schroeder]

Hi Prabhu-

My opinion: if you've tried something, tested it locally the best you can, described it on the list, and given it enough time to age, and there are no objections, check it in. If the dashboard croaks, then you know for sure :-) We can always back out.

(Oh-oh, I feel the urge to climb up on a soapbox...
I think part of the confusion is due to the fact that are multiple levels of importance to the system. And depending on the importance, different strategies are required. For example, a lex/yacc change can trash the whole system and requires more caution. Adding a new class like vtkImageConvolution has the potential to be widely used and should be discussed. Adding a class like vtkOrderedTriangulator (which I recently did) affects two other classes in the system (vtkClipVolume and vtkClipDataSet) and is very specialized. It doesn't make sense to spend the time checking with everybody else.

This is all obvious. The bottom line is that it is a process, inherently mistake prone, and requires trust to work. You do the best you can with the intention of creating something good, and expecting the help of others who share similar goals.
....off the box :-)

Will


At 11:46 PM 5/24/2001 +0530, Prabhu Ramachandran wrote:
>hi,
>
>>>>>> "WS" == Will Schroeder <will.schroeder at kitware.com> writes:
>
>    WS> Hi David- This was a class contributed by an outside party and
>    WS> sent to me. I threw it in the repository because I like to
>    WS> encourage contribution and I like to have the testing process
>    WS> chew on it to flesh out the bugs. As you can see it was
><snip>
>    WS> extreme programming process more (meaning poke something and
>    WS> see what emerges) rather than the cathedral approach of up
>    WS> front agreement on everything, which I know introduce chaos at
>    WS> time. I agree that certain additions need more discussion as
>    WS> this one did. I will definitely keep this in mind in the
>    WS> future.
>
>Aha! Gotcha!! ;) Does this mean that I can go ahead and commit my
>patched vtkParse.y and vtkParse.tab.c into CVS and let the testing
>process flesh out the bugs?  I have waited for response for almost a
>month now.  My original message is here:
>
>http://public.kitware.com/pipermail/vtk-developers/2001-April/000366.html
>
>This message mentioned that the current vtkParse.y (that is in CVS) is
>not parsed properly under bison 1.28 and that only older bison
>versions work okay.  I dont know how to fix that bug but for my
>changes (related to the Get/Set##name##MinValue/MaxValue methods) I
>patched the vtkParse.y and used an older bison and things work okay.
>
>The patch to vtkParse.y is here:
>
>http://public.kitware.com/pipermail/vtk-developers/2001-April/000368.html
>
>I spent quite some time getting all this going and dont want to throw
>all that effort down the drain.
>
>The reason why I still hesitate to commit this is that (1) I am not
>very familiar with yacc/bison and my changes are definitely hacks
>(they do work fine but...)  (2) I heard that changes to vtkParse.y are
>supposed to be done carefully.
>
>However, I have been using my vtkParse.y all this while and have had
>no problems whatsoever.
>
>Also, if the yacc/bison gurus are _really_ busy then what time is a
>good time for me to remind them?  Or are you ignoring all this because
>you are planning on doing away with all this yacc business by using
>gcc-xml??
>
>
>thanks,
>prabhu
>
>_______________________________________________
>vtk-developers mailing list
>vtk-developers at public.kitware.com
>http://public.kitware.com/mailman/listinfo/vtk-developers