[PECL-PROPOSAL] stats

Hi,
few months ago I developed a wrapper around 2 scientific libraries.
DCDFLIB (Library of C routines for Cumulative Distributions Functions,
Inverses, and Other parameters)(by Barry Brown and James Lavato)
and RANDLIB (by Barry Brown, James Lavato and Kathy Russell).
I was asked to create the wrapper by a Mr. Ed Luschei from University
of Wisconsin - Madison WI. He is a weed ecologist. The site of the
"Weed ecology" group is :
http://weedecology.wisc.edu/ . The wrapper is available here :
http://www.hristov.com/andrey/projects/php_stuff/stats/
and can be downloaded from this url :
http://www.hristov.com/andrey/projects/php_stuff/stats/stats.tgz
There is previous an effort to wrap the library by 2 scientist of UCLA :
Arno Ouwehand and Jan de Leeuw . Their implementation however is worse
since they do not validate the input parameters. In their code there
are available numerous problems that may lead to segmentation fault.
Thus I decided to rewrite the wrapper and use different function names
according to PHP's coding convention for function naming.
The documentation of the 2 UCLA's scientist can be used as reference :
http://www.jstatsoft.org/v02/i01/statistics.pdf .
I propose this wrapper to be added to PECL.

Any comments/suggestions are welcome!

Thanks,
Andrey Hristov

P.S.
AFAIK the two wrapped libs are in Public Domain therefore there is
not license problem.

Moriyoshi,
so maybe I should modify the libs and make them not to use static vars
in the functions?

Andrey

----- Original Message -----
From: "Moriyoshi Koizumi" moriyoshi@at.wakwak.com
To: "Andrey Hristov" andrey@php.net
Cc: internals@lists.php.net
Sent: Monday, August 11, 2003 10:08 PM
Subject: Re: [PHP-DEV] [PECL-PROPOSAL] stats

As I can imagine some examples that demonstrate dynamic graph drawing with
ming or gd, I think it'd be cool to have.

However, I looked at dcdflib.c and only found it was coded not in C but in
pseudo Fortran. That's worth a glance :) So the last concern is its
stability, since it simply uses static variables in the application-level
context (not in a thread-local context). This is fixable though.

Moriyoshi

"Andrey Hristov" andrey@php.net wrote:

Hi,
few months ago I developed a wrapper around 2 scientific libraries.
DCDFLIB (Library of C routines for Cumulative Distributions Functions,
Inverses, and Other parameters)(by Barry Brown and James Lavato)
and RANDLIB (by Barry Brown, James Lavato and Kathy Russell).
I was asked to create the wrapper by a Mr. Ed Luschei from University
of Wisconsin - Madison WI. He is a weed ecologist. The site of the
"Weed ecology" group is :
http://weedecology.wisc.edu/ . The wrapper is available here :
http://www.hristov.com/andrey/projects/php_stuff/stats/
and can be downloaded from this url :
http://www.hristov.com/andrey/projects/php_stuff/stats/stats.tgz
There is previous an effort to wrap the library by 2 scientist of UCLA :
Arno Ouwehand and Jan de Leeuw . Their implementation however is worse
since they do not validate the input parameters. In their code there
are available numerous problems that may lead to segmentation fault.
Thus I decided to rewrite the wrapper and use different function names
according to PHP's coding convention for function naming.
The documentation of the 2 UCLA's scientist can be used as reference :
http://www.jstatsoft.org/v02/i01/statistics.pdf .
I propose this wrapper to be added to PECL.

Any comments/suggestions are welcome!

Thanks,
Andrey Hristov

Moriyoshi,
I forgot to say that examples most probably will be available soon
written by Ed Luschei. He uses GD for visualisation.

Andrey

----- Original Message -----
From: "Moriyoshi Koizumi" moriyoshi@at.wakwak.com
To: "Andrey Hristov" andrey@php.net
Cc: internals@lists.php.net
Sent: Monday, August 11, 2003 10:08 PM
Subject: Re: [PHP-DEV] [PECL-PROPOSAL] stats

As I can imagine some examples that demonstrate dynamic graph drawing with
ming or gd, I think it'd be cool to have.

However, I looked at dcdflib.c and only found it was coded not in C but in
pseudo Fortran. That's worth a glance :) So the last concern is its
stability, since it simply uses static variables in the application-level
context (not in a thread-local context). This is fixable though.

Moriyoshi

"Andrey Hristov" andrey@php.net wrote:

Hi,
few months ago I developed a wrapper around 2 scientific libraries.
DCDFLIB (Library of C routines for Cumulative Distributions Functions,
Inverses, and Other parameters)(by Barry Brown and James Lavato)
and RANDLIB (by Barry Brown, James Lavato and Kathy Russell).
I was asked to create the wrapper by a Mr. Ed Luschei from University
of Wisconsin - Madison WI. He is a weed ecologist. The site of the
"Weed ecology" group is :
http://weedecology.wisc.edu/ . The wrapper is available here :
http://www.hristov.com/andrey/projects/php_stuff/stats/
and can be downloaded from this url :
http://www.hristov.com/andrey/projects/php_stuff/stats/stats.tgz
There is previous an effort to wrap the library by 2 scientist of UCLA :
Arno Ouwehand and Jan de Leeuw . Their implementation however is worse
since they do not validate the input parameters. In their code there
are available numerous problems that may lead to segmentation fault.
Thus I decided to rewrite the wrapper and use different function names
according to PHP's coding convention for function naming.
The documentation of the 2 UCLA's scientist can be used as reference :
http://www.jstatsoft.org/v02/i01/statistics.pdf .
I propose this wrapper to be added to PECL.

Any comments/suggestions are welcome!

Thanks,
Andrey Hristov