Re: sccons: what's happening?

[Steven Knight]

> "The design competitions have all been concluded. CodeSourcery is actively
> developing the testing tool (QMTest) and issue-tracking tool (QMTrack).
> Pre-release versions of these tools will be available sometime in the summer
> of 2001. At this time, no implementation is being done on the configuration
> and build tools."
>
> As I understood it the winner of the design competition was going to be paid
> to develop his proposal.
>
> Steve: any news on what's happening with that?

Okay, I guess I can start going public here.  Here's the summary:

Although they didn't fund an implementation, Software Carpentry did
contract me for another round of design work on ScCons.  I tried
off-and-on for several months, but just couldn't muster the enthusiasm
to generate Yet Another design document...

So around about January, I decided I was just going to go ahead and
implement it on my own, and effectively resigned from the effort.
(Since I hadn't produced anything tangible towards what was contracted
for, I didn't take any Software Carpentry money for anything after the
contest itself.)

So on the side, I'm working on developing Cons-in-Python as an
independent effort.  I've renamed it Scons (Software CONStruction tool)
to distinguish it slightly from Software Carpentry, still honor its
roots, and to make it a teeny bit easier to type.  It's still a ways
from release, but I have a good amount of the infrastructure in place
(including, of course, an extensive testing infrastructure... :-). It
will currently build files, but I haven't yet coded up the dependencies
or .consign stuff; they're the next steps, followed by -j support, which
is going to be present from the beginning.

The design feels pretty clean, owing mainly to having Cons classic as
a reference for how to do most things, and how not to do some others.
My Python learning curve has been steeper than I envisioned, I imagine
mainly because I'm trying to do more intricate things right away than
is usual for a Python novice.  I don't care for the organization of
the Python documentation.  I've spent a *lot* of time searching for
things in the library only to find them in the reference section or the
tutorial, or vice versa.

Enough of the Scons infrastructure is there that I can probably, at
this point, profitably bring on board a few people who want to help
with development.  If anyone out there is interested, let me know
(particularly if you know Python well...).

> As much as I love the Perl based Cons I thought that a well designed suite
> of tools was a nice idea.

I agree.  There's no reason why Scons couldn't be part of it, if they
ever get to that point.  But I honestly don't know where Software
Carpentry is with the other parts of the effort, and as you point out,
they're not being very informative about their progress.

        --SK