Re: private functions in the core Octave distribution

[John W. Eaton]

On  1-Dec-2009, S ren Hauberg wrote:

| tir, 01 12 2009 kl. 16:17 -0500, skrev John W. Eaton:
| > I checked in the following change to make some internal plotting
| > functions private or to convert them to subfunctions (if possible).
| 
| Nice!

OK, I checked in another batch.  In this set, I moved some tests from
the private functions to the calling functions since we don't have a
way to test private functions directly.  I also found and removed some
redundant tests (the tests in __quantile__ appeared to be a subset of
the tests for quantile).  The only case I'm not sure about is
__xzip__, which has some tests that I don't think can be rewritten in
terms of the functions that use the private __xzip__ function.  For
now I've commented them out, but maybe it would also be OK to remove
the tests that can't be rewritten since they are for cases that are
not supposed to happen, and we should be able to guarantee that all
calls to private functions will be valid.

I was about to make the __additional_help_message__, __makeinfo__, and
__strip_html_tags__ functions from the scripts/help directory private,
but then noticed that some Octave Forge scripts call __makeinfo__
directly.  

As far as I can tell, __additional_help_message__ is only used in
Octave's help and print_usage functions, so I made it private.

I don't see any uses of __strip_html_tags__, so maybe that should be
deleted from Octave?

jwe