[qmtest] RFE concerning 'Extension' class

[Mark Mitchell]

> In other words: the question is not how to continuously support classes
> that contain an 'arguments' class variable, but how to make it possible
> to avoid such classes in the future.

Assuming that's a major problem...

I fully understand why this stuff is cool -- but I'm not quite sure what
problem it's solving.

Backing up, let me be very clear about our agenda here, which may of
course not be the same as yours. :-)

(1) Do not break things for existing customers and users -- they *hate*
that and we hate life when customers hate us. :-)

(2) Enlarge the total QMTest user-base.

I understand that (2) is helped by making things easier and cleaner to
program.  I totally get that.

However, I believe that the key thing that we could for (2) would be to
enhance the directly user-visible functionality.  For example, if we
provided a test class that made it very easy to test web applications,
maybe that would drive adoption into the web application area.  Or,
maybe -- as you suggested -- polishing up our compiler testing test
classes.  Or, GUI enhancements, such as creating a directory/file style
dialog box for selecting tests and resources and for creating new test
directories.

The truth is that if you write some cool code that makes the guts of
QMTest cleaner, I'm probably going to review it and we're probably going
to incorporate it.  That's all well and good.

But, really, I'd like to stay out of the guts as much as possible,
except for bug-fixing.  I'm much more excited about user-visible changes
that I am about tweaking the internal API.

-- 
Mark Mitchell <mark at codesourcery.com>
CodeSourcery, LLC