[MLton] new SML project?

Vesa Karvonen vesa.karvonen@cs.helsinki.fi
Wed, 22 Jun 2005 12:07:29 +0300

Quoting Stephen Weeks <sweeks@sweeks.com>:
> I wonder if we might be able to find a project of broad enough interest
> to people that we could start development of a new (non-compiler!) piece
> of software written in SML, which naturally would need libraries. The
> software should have the following properties:

Another option would be to take an existing program written in SML and
refactor it. Undocumented, unpolished, ad-hoc utilities in the program
code would be turned into separate documented and polished libraries and
the program code would be changed to use the separate libraries.

This could have several valuable benefits. For one thing, this might, *in
the long run*, reduce the maintenance burden of the refactored program.
The externalized libraries could be maintained (documented, debugged,
enhanced, ported, ...) by developers other than the developers of the
original program. Future libraries required by the program could also be
externalized in the same fashion. The architecture of the original program
might become clearer. The created libraries would already be used in at
least one application. This is likely to motivate the maintainers of the
libraries. Feedback from one set of users that care is better than
feedback from zero users.

The main downside with this approach is that it would obviously initially
require close collaboration from the developers of the original program.
It would most likely take away resources from adding new features to the
original program for some time.