Re: [OctDev] Question on performance, coding style and competitive software

[Jaroslav Hajek]

On Fri, Apr 24, 2009 at 8:47 PM, Przemek Klosowski
<address@hidden> wrote:
>   Users running a package on Matlab IMHO bring little to Octave
>   development, as Octave is not involved anywhere. The benefit goes to
>   the package, but it is up to the author to consider whether it
>   outweighs the burden.
>
> There is a significant second-order positive effect for Octave:
> M/O-compatible packages get in the 'widely-used->rapidly-improving'
> feedback loop, which then helps the Octave's loop.

But to "help the Octave loop" the users need to run the package on
Octave, not Matlab. Those who use it on Matlab only contribute to the
package's quality. This may still be good for Octave in the long term,
I wasn't questioning that.
Nevertheless, we don't want Octave to be a mere follower of Matlab. If
there's no code using unwind_protect, then unwind_protect will be
always buggy. That's why using Octavish features helps Octave as well.
If users avoid using any such features, we may just not develop any.
But many of us, including me, do use and develop Octave because it can
offer them something Matlab doesn't (and they have no option to
improve Matlab, that's actually illegal due to EULA), besides the
freedom.
So I want to use default function arguments, I want unwind_protect, I
want oct-files, because I value these features, even though they're
not in Matlab.

> Note how all successful languages have a wealth of libraries
> ('batteries included', i.e. a rich vertical application environment a
> la CPAN, Beans, Python libs, PHP Pear)---it's a chicken and egg
> situation, which Octave always leveraged through its other leading
> product ties.

I'm not sure how much this relates to our dispute.
We certainly want a lot of libraries for Octave. But programmers
started to write libraries for Python because of Python's quality, not
of its compatibility with something.

-- 
RNDr. Jaroslav Hajek
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz