Octave 3.0.1 bug_report fails

[Michael Goffioul]

On Fri, Jul 11, 2008 at 9:06 PM, John W. Eaton <jwe at bevo.che.wisc.edu> wrote:
> OK, now we have separate versions of the mkoctfile, octave-config, and
> octave-bug scripts/programs on Unix and Windows systems.  I think this
> is a potential source for confusion because the separate versions are
> likely to behave in slightly different ways and/or diverge over time.
>
> Would it be better to simply use the C++ versions on all systems?
>
> Is there any good reason to not do that?

The nice thing about the script version is that it's rather easy to
maintain, does not require compilation, can make use of all
standard UNIX tools (sed, grep...) and is UNIX-cross-platform.

The problem is that they're unusable under Windows
(even on MinGW-compiled octave). The 3 C++ version I wrote
however are quite different:

- mkoctfile.cc is a real one-to-one translation of the script
version; this one was mandatory to have under Windows
in order to compile octave-forge packages for instance

- octave-config.cc is there for "symmetry", but I actually
never used it

- octave-bug.cc takes a radically different approach than
the script version; this one is really Windows specific,
although it could also include a one-to-one translation
of the script (stream editor + mail)

> If so, then who will take on
> the job of ensuring that the behavior of the separate versions of
> these programs are equivalent?

Those scripts actually change very rarely. Maintaining
equivalence does not seem to be a tough job. Currently,
when I detect a change has been made on mkoctfile,
I port it to my C++ version, but I think it only happened
once or twice. However, it may happen that I miss some
changes, so if both script and C++ version are going to
co-exist, it would be nice that if anybody makes a change
to one version, either he also port the change to the other
version or notifies the mailing list about the change and
ask for someone (for instance me) to port it.

Michael.