Re: FYI [Was: libevaluator library and GSL]

["Aleksandar B. Samardzic" <asamardzic at matf dot bg dot ac dot yu>]

On Mon, Sep 22, 2003 at 11:55:04AM +0100, Alberto Manuel Brand?o Sim?es wrote:
> Is at this point that people like me, with a built-in parser and
> evaluator (on numexp -- numexp.sf.net) should or not try to move to the
> new library and contribute.
> 
> Aleksandar, can you give a quick sumary about what libmatheval supports
> at the moment?
> 
> See you
> Alberto Sim?es
> 
> On Mon, 2003-09-22 at 11:40, Brian Gough wrote:
> > Aleksandar B. Samardzic writes:
> >  > Libevaluator library (discussed here two weeks ago) is now part of GNU
> >  > project.  During conversion, a name change is requested, so library is
> >  > called libmatheval now and its home page is at:
> >  >   http://www.gnu.org/software/libmatheval/
> > 
> > Cool.  I prefer the new name.
> > 
> > I will send you some patches.
> > 
> > One thing I'd recommend is using a C program to run the tests (as in
> > GSL) rather than requiring Guile, since not everyone has that
> > installed (even though they should ;-) ).

Hello all.

I'll reply to both Alberto and Brian.

Presently, libmatheval supports rather simple expressions: basic
arithmetic operators, basic mathematical functions (exp, ln, sqrt, sin,
cos, tan, ctan, inverse sin, cos, tan and ctan, hyperbolic and inverse
hyperbolic sin, cos, tan and ctan and finally abs), decimal (I mean base
10) constants and arbitrary variables (whatever is found by parser
described by [_a-zA-Z][[_a-zA-Z0-9]* regex is considered variable).
Library is able to parse expression, to evaluate expression for given
variable expression and to calculate symbolic derivative over given
variable.  I'm of course ready to work on extending this functionality,
so - Alberto, please go on with what would be necessary to add to
libmatheval in order to have some convergence between Numexp and
libmatheval development (if you think that this would be appropriate at
all).

Btw, I was contacted in the meantime by several Numexp fans; they
objected duplicated effort.  They certainly have a point, but
unfortunately I wasn't aware of Numexp existence back in 1999 when I
first wrote libmatheval.  And now when GNU people decided such a simple
library like libmatheval could be worthwhile to exist separately (I
believe Brian words from previous thread that I passed them helped to
create such opinion), I guess I have no other option than to keep
maintaining it.  But still - maybe we could try to work together.

Any suggestions and patches for libmatheval are more than welcome. I'll
need some more time to complete libmatheval setup in sense of being part
of GNU project (legal papers exchange with FSF before all).  I've just
created bug-libmatheval@gnu.org mailing list, so if anyone interested to
help in libmatheval development - please let's continue discussion
there, in order to not disturb other gsl-discuss subscribers.

Regarding Guile - I just decided to strictly follow GNU coding
standards.  And I guess I knew this will also help a bit to libmatheval
to become part of GNU project :)

Regards,
Alex