Re: dirac(x) and heaviside(x)

[Søren Hauberg]

tir, 07 12 2010 kl. 13:17 -0600, skrev Jordi Gutiérrez Hermoso:
> 2010/12/7 John W. Eaton <address@hidden>:
> > Currently this just prints a special message for core Matlab
> > functions.  Extending it for package functions would not be hard, but
> > who maintains the list of functions?
> 
> Soren, I guess? The list of available functions does seem to be
> maintained. It's that link to the alphabetical list of functions I
> provided earlier. The problem is that it's currently disconnected from
> the core Octave distribution.

Oh, please don't put such things on my shoulders... Carlo and myself
spend quite some time on making Octave-Forge as self-maintained as
possible simply because I couldn't be counted on to do proper
maintenance.

You might be able to download the list of functions from the
Octave-Forge server form Octave. I mean, we could let

  pkg -forge rehash

(or something like that) download a list of packages, functions and
missing-hooks from the Octave-Forge server. This command could be
executed during Octave install (or when new Octave releases were made)
to ensure some initial list.

> Weren't 'Forge releases in the past made to coincide with Octave
> releases, in the Paul Kienzle days? I remember something like this.

Yeah, we did something like that when we had monolithic releases. As
somebody who actually did several monolithic releases, I can, however,
say that this process was just pure pain.

> > Do Matlab users typically not know that a function they are using
> > comes from a Matlab Toolbox(TM) instead of Matlab?
> 
> I don't know about this.
> 
> We do get many Octave users coming from Matlab, and they're very often
> students who are trying to run code that their instructor gave them.
> They grab said code, they try it in Octave, they find some function
> that isn't there, and then confusion happens. Often they use Matlab in
> computer labs where all of the toolboxes necessary to run the
> instructor's code is installed. We seem to get plenty of these cases
> in the help mailing list, and I've observed them in #octave too.

Yeah this is my impression as well. Many Matlab users don't actually pay
for the software themselves so they don't really know what was bought.
As such, they don't really care about toolboxes.

> Perhaps if the missing.html page just mentioned Octave-Forge with the
> same link I provide to the alphabetical list of functions? This could
> be easy and it would still allow for the decoupling of Octave and
> 'Forge. Maybe this decoupling is desirable in order to let package
> authors make releases more frequently than Octave.

Currently, we don't actually have the missing.html page. I recently sent
a proposal for such a page to the maintainers list. I think I included
the link you mention, but let me know if you think the text could be
improved.

Soren