[Insight-developers] RFC: Agressive error checking during compilation?

[Daniel J. Blezek, Ph.D.]

Paul, et. al.,

  We've been working on Nightly builds for Insight to address exactly
these problems.  The URL is:
http://public.kitware.com/Insight/Testing/HTML/TestingResults/Dashboard/MostRecentResults-Nightly/Dashboard.html

  Currently we have builds occuring on a Sun, Linux and an SGI (which has
been having problems).  We also report changes made in the last day, and
coverage information.  The basic testing framework is outlined below.

We have two classes of build/test combinations: Nightly and Experimental
  - Nightly builds are all done w/the same code, a CVS snapshot at a
particular time
  - Experimental builds can be done anytime during the day, and will
report code that is modified or added from the Nightly baseline
  - Any user of Insight may submit Nightly or Experimental builds to the
Kitware site, please contact me if you are interested (currently Unix
only, until some CMake issues get resolved)


Dashboards are generated every 30 minutes through out the day containing
the Nightly builds as well as any experimental builds that have happened
during the day.  We are shooting for having as many platforms as possible
contribute to the nightly builds, and anyone that is interested, to
contribute to the Experimental builds.  One of the Experimental builds
will likely look for CVS commits, and compile code as it comes in, giving
you a look at how your changes do on other platforms.  All of the builds
have the -Wall flag on, so we get a count of as many warnings as possible.

  I've been away for 2 weeks, but will be working on the testing
infrastructure through the end of the year, and we'll also add a bit
better navigation to the testing results.

Take care,
-dan


On Tue, 28 Nov 2000, Paul Hughett wrote:

> 
> I just fixed a bug in one of my programs in the repository that
> involved a misspelled variable name.  The reason that I bring this up
> is that this bug would never have made it past the first compilation
> in one of my own projects, because I routinely compile with the
> optional warnings turned on (gcc -Wall) and the compiler would have
> told me there was a problem.  Insight, however, specifies only the
> default warnings for gcc.
> 
> Perhaps we should enable the optional warnings for the compilers we
> use?  I've found the warnings provided by gcc to be very useful in
> catching various bugs that would be much more expensive to debug by
> hand.  Does anyone think this is a bad idea?
> 
> Paul Hughett
> 
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
> 

--
Daniel Blezek, Ph.D.
blezek@crd.ge.com
Visual Information Program
Electronic Systems Lab
GE Corporate Research & Development