Re: I resign as Guile maintainer

[Greg Harvey <Greg dot Harvey at thezone dot net>]

mstachow@alum.mit.edu writes:

> I can volunteer to do it. 

(big sigh of relief from everyone (including myself) who thought there
was a remote possibility it would end up being me ;). This is really
great news.

> However, if someone who is qualified and can devote more time than this
> is available, I would defer in favor of them.
> 
> If offered the position, I expect to spend most of the time integrating
> the code
> of others, promoting Guile to the community and trying to address
> perceived
> and real shortcomings, rather than launching ambitious new development
> projects
> myself; 
>
> I think right now Guile mainly needs to be seen as more widely
> usable
> more than it needs any particular feature, and there are already many
> people
> working on most of the key features that are needed long-term.
> 
> I think the most important specific priorities right now are
> documentation
> (first a thorough and complete reference manual, then a user's guide),
> and the perceived performance problems, which are mainly caused by the
> startup time. These are the two top issues cited as reasons for not
> using
> Guile as the (or one of the) extension language(s) for a project.

I think that, on the performance side, there are a few `quick fixes',
precompiling whatever's possible in boot-9 with hobbit or moving
functions out of boot-9 to c code by hand. I think that a combination
of the two is probably best, since there are some things in boot-9
that you'd really like to use from c code. A little of this can go a
long way: back when I was making thing releases, startup was a little
over twice as fast as the guile at the time... there were some fatal
bugs, of course, but hey, you didn't have to wait too long for it to
crash :). I realize the objections to these, but I think that the
benefits (you can start guile on a 486 and not have to find something
else to do while it loads) far outweigh the disadvantages. A lot of
better solutions have been proposed (guile binary format files, for
example). These best solutions will take time (possibly a very long
time; godot, anybody?... or the generational garbage collector >;'); I
think that guile has lost a lot of potential users by searching for
the perfect solution, or, more specifically, the *implementation* of
the perfect solution; just about everybody can describe the perfect
solution ;). It makes interesting reading. It usually does very little
for guile. Adding a temporary solution that works really doesn't make
it any harder to add a better solution in the future, and may actually
suggest ways of accomplishing the better solution (it could be argued
that a temporary solution will mean that there's less of a chance that
someone will implement the better one; personally, I don't think that
holds too much water; if having a kludge in doesn't make you feel like
working towards removing that kludge, you probably weren't going to do
it anyway).

Cool, a rant :).
-- 
Greg