[Gardeners] Random Thoughts

[Jean-François Brouillet]

If we approach this "gardening" task with some amount
of humility, it seems that there are quite a few areas
where we could improve things, none of which very glamorous,
but that would definitely help "the cause" :)

- making things *work* first time.

I know about Peter's LispBox package and I believe it to
be a great start, but it seems to me that other areas could
be improved as well.

With all those free (and non free) implementations around,
what can *we* do to make the install experience as painless
as possible for the "newbie" ?

It doesn't seem reasonable to pick a given implementation
at the expense of every other. What we could do, is bridge
the various gaps, in one way or another, so that whatever
implementation is chosen, we could offer a kind of guarantee
that:

- things will install first time. "Buy in 1 click!" :)
- asdf will be made part of the box, *and working* without
   having to follow the otherwise excellent Edi Weiz tutorial

- try to iron out some/most/all of the "non portability"
   problems in the non ANSI part of the CL feature set,
   by providing higher level wrappers where possible (threads,
   sockets, GUI jump to mind)

Ideally, we could come up with plenty of "small" projects
with only a handful of each of us in each, and as each such
project grows in scope/complexity, divide it again into
smaller chunks, again for a new team to pick up.

I don't see it as a problem if some commercial company
(Allegro, Xanalys, Digitool) have to gain some through
our work, in much the same way I wouldn't anticipate
any of them starting whining because we're killing the
differentiators.

In my mind. there should be just *one* CL, no matter
where one gets it from (Commercial or free), and that
CL should become as ubiquitous as Java _is_. Whether
people would, en masse, move to CL is _their_ decision,
whether we should prepare for such a move, is _ours_.

--
Jean-François Brouillet
verec at mac.com