Re: [Gcl-devel] 2.6.2

[Vadim V. Zhytnikov]

Camm Maguire ?????:

> Greetings!  OK, I've just committed the readline patch.  I'm doing the
> full test suite now, but barring the unforeseen, I'd like to tag this
> as 2.6.2, release, and then move on to head.  Please speak now if
> there are problems.  On machines to which I have no access, I'm
> assuming that nothing that was working has been broken.  Please
> correct me if this is not the case.  We might want to make the
> use-fast-links nil switch on Windows automatic i.e. handled by
> configure, but apart from this I don't know of anything else that
> needs to go into stable.
>
> I'll report back on my test results when done.  Basically
> self-compile, ansi-tests, Paul's compiler tests, maxima, acl2, axiom,
> nqthm-1992, pc-nqthm-1992, and builds on all 12 Debian arches is my
> test suite.  Mac-OSX, BSD, and/or Windows might want to consider the
> same.  It will be a few days before all this is completed.  Needless
> to say, please don't commit anything non-platform specific to the
> stable branch in this period.
>
> (BTW, readline is now initialized automatically.  Completion can be
> customized somewhat via the variable si::*readline-prefix*.  See the
> info pages for details.  The Debian maxima package will be setting
> this var to "maxima::$" for now.)
>
> Take care,

I observe a few problems on mingw:

1). mingw.defs defines GCL_HTML in old style (gcl_toc.html etc)
thus build on mingw fails on info -> html conversion.
But even if this problem is fixed make install fro .html
fails on mingw and need to be updated according to new
html scheme in GCL.

2) It is good idea to mention in readme.mingw:

  2.1)  GCL build on mingw requires not only
  mingw+msys+msysDTK but also additional
  texi2html script (with !#/usr/bin/perl -> !#/bin/perl).

  2.2)  Mention that readline doesn't work on mingw.
  Although with some modification GCL can be compiled on
  mingw with readline (remove all references to
  ncurses in configure) but result is broken.

Finally, what about including into gcl 2.6.2 some memory
allocation enhancements?

3). More aggressive default allocation-growth parameters.

4). Make (si::set-gmp-allocate-relocatable t) default.

5). Increase a bit initial hole size.



--
     Vadim V. Zhytnikov

      <address@hidden>
     <address@hidden>