Re: pytave: adding exec and eval commands

[Mike Miller]

On Fri, Mar 11, 2016 at 00:26:10 -0800, Colin Macdonald wrote:
> On 10/03/16 12:02, Mike Miller wrote:
> > FYI, I have a clone of pytave on bitbucket, would you consider making
> > yours a fork of mine and rebasing your changes on that? You can then
> > submit PRs there if you like that interface. 
> 
> Ok, I hope I've done so...

I don't know if you saw my comments on bitbucket, but it would be better
to use bookmarks than branches for commits intended for the default
branch. Mercurial bookmarks are more equivalent to git feature branches.

> > The direction I think makes sense *eventually* is to treat all Python
> > objects as first class objects in Octave, whether they're "things" or
> > "callables", versus the quick hacks I was playing with last year to
> > evaluate objects vs function calls.
> 
> Sounds good to me: I can do `py.exec(str)` where `str` is a long string
> of python code.
> 
> (Future code in Symbolic might be written differently but for now its
> just strings of python code).

I've applied your changes to pytave now. You can strip the feature
branches from your repository, look under settings.

I took out the dlopen hack, please try building again and see if it
works now. If not, what is the value of PYTHON_LIBS in your Makefile?

I've spent a good amount of time cleaning up and fixing some things in
the pytave repository over the past couple of days, looking forward to
improving it a lot this year.

Since your package is probably the biggest driver of these improvements
at this point, I'm looking forward to your ideas or contributions to
help make it work better with Octave. For example automatic importing of
packages, as in the py function, does not work with pyeval yet.

-- 
mike