Re: [Texmacs-dev] Disastrous boot time for new versions

[Joris van der Hoeven]

Hi,

Thanks for your detailed compairison of Scheme implementations.

> In my humble opinion, MzScheme
> http://www.plt-scheme.org/software/mzscheme/ is the most complete,
> best documented and very actively supported implementation. It has
> been around for quite some time and many people are using it. MzScheme
> has nice FFI very well suited for writing extensions. Moreover, SWIG
> officially supports mzscheme as one of its interface languages
> (http://www.swig.org/). Mzscheme is R5RS compliant and comes with
> module system, namespaces, OO, neat resource managment (custodians),
> units, hygienic macros etc. Old style define-macro is also supported
> for backwards compatibility. Very good documentation is an important
> strength. MzScheme is being developed by somputer scientists
> originally at the Rice University, TX, but now in Brown, North Eastern
> Universities, University of Chicago, Univ. of Uta, etc. These guys are
> writing scientific papers presumably with math formulas, trees, data
> structures, etc. They could be interested in TeXmacs per se and might
> help with the port:-)

I took a look and this seems indeed to be a quite cool implementation:
I measured it to be five times faster than guile on the computation
of the Fibonacci number (fib 32) using the stupid exponential algorithm,
so making a move to this language might indeed reduce our problems with
boot times quite a lot. Moreover, they support a guile compatability library.
I hope that the version will match with the one we use...