Re: Missing functions in the statistics package.

[Mike Miller]

On Mon, Sep 14, 2015 at 09:49:46 -0700, Dag wrote:
> I cloned the source of statistics as follows: "pkg install -forge -local
> statistics".

This is an install of the package, not the source. To clone the source
repo for statistics,

  hg clone http://hg.code.sf.net/p/octave/statistics

Do you have any thoughts about how we could make that step more discoverable?

> Not sure how that works.

With the full cloned source you should now see the make targets Oliver
was describing.

> I was hoping of being able to use many installations of the same program. If
> they are siloed, they will ideally not interfere with each other. (I am a
> relative newbie, still learning.) For example, you can have several
> installations of gcc and g++ installed from repositories without conflict.
>
> Currently I have the installed Octave 3.8.1 from the Linux Mint 17.1
> repositories, based on Ubuntu 14.04. And then also a local installation of
> v. 4.1.0+ from the Mercurial repository. I can control which installation to
> launch by using soft links linked to binaries.

Note that Octave adds "-version" suffixes to all of its executables.
The way I work is keeping the "octave" symlink pointed at the latest,
but pass OCTAVE=octave-3.8.2 and MKOCTFILE=mkoctfile-3.8.2 variables
when I want to test against older versions.

> I hope to write C/C++/Fortran source files later, and that requires more
> skill than only editing m files, because you have to recompile and perhaps
> reinstall packages before testing.

I hope you can see when working with the cloned source repo that you
can just use "make run" as Oliver described to automatically compile
and run an Octave session pointing to your local changes.

Hope that clears things up a bit,

-- 
mike