|
|
Ublas : |
Subject: Re: [ublas] Snapshot 20081116
From: Rutger ter Borg (rutger_at_[hidden])
Date: 2008-11-23 05:17:59
Dear Thomas,
Thomas Klimpel wrote:
> Do you mean you would be interested to participate in an effort to pitch
> the numeric bindings library as a Boost Library on its own? I guess your
> contributions would be most welcome :-)
Exactly. I am interested in helping. I've using it since one of the earliest
releases, I've written a library around it (see
http://www.terborg.net/research/kml/), and I've have been promoting the use
ever since.
> Or do you just want to know about the plans to move forward with the
> numeric bindings library? I guess the long term plan is indeed to get the
> numeric bindings library accepted as a "true" boost library. For the short
> term, I think it is more important to address the "backlog", i.e. review
> and integrate existing user patches (and add tests for the corresponding
> functionality) and add badly missing functionality (like full support for
> 64-bit platforms).
Ideally, we could decide on where the bar is for submission for review, so
we can work towards that. I think if we could get it up to the quality to
get it accepted into Boost, its user base could grow and thereby increasing
the speed of development.
Is there an overview or bucket that contains those patches and/or TODO list?
Would it be an idea to use Boost's trac wiki/ticketing facilities for this?
> Another direction of development concerns semi-automatic generation of the
> source-code for the bindings. The attached python-script tries to
> interpret fortran source files from lapack-3.1.1/SRC, in order to generate
> some of the required boiler-plate code. The script is still far from being
> able to parse every file in lapack-3.1.1/SRC, but someday... It is also
> not yet able to generate template code, but having the prototypes and
> inline functions is also not bad. And someday...
Cool, I'll look into it.
> I don't quite follow what you want to say here. It's no surprise that some
> users of uBLAS will need functionality like eigenvector computations, and
> will therefore be forced to use the numeric bindings library. The numeric
> bindings library on the other hand is most often used via uBLAS, so it's
> only natural that it uses the ublas mailing list.
Sure, I understand. I was merely making the non-scientific observation that
a non-neglegible part of the posts on this list concern the numeric
bindings library. Also, I think it will be great if the enormous effort of,
e.g., LAPACK, ATLAS, and others like it, doesn't have to be replicated in
uBLAS before it can be used.
I think the numeric bindings will work with any vector and/or matrix type
that adheres to certain concepts (e.g., like those defined by uBLAS),
doesn't it? In other words, making it an independent library is something
that's not infeasible.
Kind regards,
Rutger