On Fri, 2004-10-15 at 12:00, Raymond Toy wrote:
> Since we're making some pretty drastic changes with the case
> sensitivity changes now, what other drastic changes are allowed?
I don't want to see too many drastic changes. However, I wouldn't call
most of what you listed below "drastic." I don't know about everybody
else, but the code downcasing (and indenting) made a huge readability
difference to me. A few more changes in similar spirit would be welcome.
> Some things I'd like to see, not necessarily now:
>
> o Trash old code like stuff for pdp-11, TI explorer, etc.
I don't think we should be too afraid to make these changes. Some of the
old code really gets in the way. It isn't doing anything now. Anyone
interested in history research can look at older versions of Maxima in
cvs.
I would say that anyone who feels like spending the time to carefully
remove various #+- sections should do so. However, we should make a
clear list of which *features* are definitely obsolete first.
> o Comment out (or remove completely?) multiple definitions of the same
> function. Then see what breaks and then fix it.
> o Same with macros.
We have already done some of that with the patches from Andreas Eder.
Andreas has indicated his desire to continue with his work. I say more
power to him. If a developer feels motivated to do this sort of tedious
maintenance work, I am reluctant to stop him. I have higher priorities
on my own list, however, so this sort of thing won't get done by me for
quite a while.
Incidently, let's make that "remove completely" not just "comment out".
When we are ready to remove something, it should be really removed.
> o Rename all the special vars from <var> to *<var>* when possible.
> Of course, some of these are visible to maxima, so we don't want to
> rename those, but these are all of the form $<name>, so that's ok.
> I find it very confusing that special vars aren't denoted in some
> better way in the code. I certainly can never remember which
> variables are special, and some of the variables have particularly
> annoying names like "y", so you might get really bizarre results in
> some cases if your lisp code uses y, thinking it is lexical.
>
> All of these changes (except the first) have major implications, and
> could break all kinds of code, but I think the benefits will be worth
> it.
These are all things that need to be done. I would like to see the most
trivial happen first, i.e., removing all #+pdp10 code. The more invasive
changes should probably wait for 5.9.3.
--Jim