[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Funding - Specifically identifying allocations
From: |
Paul Kienzle |
Subject: |
Re: Funding - Specifically identifying allocations |
Date: |
Thu, 29 Sep 2005 00:19:11 -0400 |
On Sep 28, 2005, at 11:35 AM, David Bateman wrote:
Søren Hauberg wrote:
Hi,
Related question: We had a plan to develop a system for delivering
add-on packages for Octave. This would complement Octave-Forge.
How is that coming along?
Well, I guess I should answer that one. The package system should be
done, but I haven't had the time to port octave-forge (I need to
learn the autotools first). So status is that I think it's done, but
it hasn't gone through the large test of porting octave-forge.
Soren,
I don't think you should port "octave-forge", just a single toolbox
would be a good proof.. The major problem I see is not autoconf
per-se, but rather the large amount of cruft for earlier versions of
octave.. The autoconf stuff is effectively there to identify the needs
of the version of octave for which octave is being built, with a very
small component there to check for particular libraries that might be
needed (cf. gmp/cln/ginac for the symbolic toolbox).. The cruft in
autoconf you can ignore, and its only the external dependencies that
need to be treated (ie we're targeting only the very latest and
probably only the 2.9.x octave tree). If you pick the signal toolbox
there are no external dependencies and so after decrufting of the
code, it might be packaged without autoconf... A good test case that
would need autoconf is the image toolbox that needs libpng and jpeg-6b
installed, and a relatively simple autoconf will suffice for that..
Once the process is in place with those two examples, the rest of
octave-forge might follow very rapidly...
David,
There are many packaging challenges in octave-forge including the ones
that you mention above.
Here are some of the issues we will encounter:
* main/signal has compiled components
* main/image has library dependencies
* main/vrml has external program dependencies
* main/audio installs a complementary binary and sample data
* main/gsl has generated code
* extra/mex and extra/engine installs headers and libs
* main/comm has texinfo documentation
* extra/graceplot installs as an 'alternative'
* extra/MacOSX is OS dependent
* extra/NaN wants to install in octave or matlab
There are many cross cutting concerns as well:
* removing cruft from old octave versions
* dependencies between packages
* categorical indexing
* automatic running of embedded tests
* PKG_ADD tags
By the time octave-forge is chunked and packaged, the install will have
pretty good coverage of the features it needs.
- Paul
-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.
Octave's home on the web: http://www.octave.org
How to fund new projects: http://www.octave.org/funding.html
Subscription information: http://www.octave.org/archive.html
-------------------------------------------------------------
- Re: speed of octave interpreter, (continued)
- Re: speed of octave interpreter, Andy Adler, 2005/09/27
- Funding - Specifically identifying allocations, Robert A. Macy, 2005/09/27
- Funding - Specifically identifying allocations, John W. Eaton, 2005/09/27
- Re: Funding - Specifically identifying allocations, Javier Arantegui Jimenez, 2005/09/28
- Re: Funding - Specifically identifying allocations, Francesco Potorti`, 2005/09/28
- Re: Funding - Specifically identifying allocations, Miquel Cabanas, 2005/09/28
- Re: Funding - Specifically identifying allocations, Bill Denney, 2005/09/28
- Re: Funding - Specifically identifying allocations, Mike Miller, 2005/09/28
- Re: Funding - Specifically identifying allocations, Søren Hauberg, 2005/09/28
- Re: Funding - Specifically identifying allocations, David Bateman, 2005/09/28
- Re: Funding - Specifically identifying allocations,
Paul Kienzle <=
Re: speed of octave interpreter, J LOVE, 2005/09/26
Re: speed of octave interpreter, Tom Holroyd, 2005/09/26