[SciPy-dev] Re: PlotCanvas

[eric jones]

> Not exactly. What I really need is to tell plt about a new plot_canvas
> it did't create by itself! Well, you could say I have to smuggle it in:)

Ahh, ok.

> The way to do that is to extend the "case" in figure() so when
> which_object is not numeric and not in _figure to check whether
>    type(which_object) == type(<instance of plot_canvas or derived class>)

> If I understand that correctly something like that is definitely not
> possible in python < 2.2 (and I am not sure it is possible in 2.2).

Your hack is fine for now I think.  I'll apply the patches.

> One comment from my side: How do you distinguish between classes/types
> and functions then? (I don't see real evidence for a verb vs. noun
> scheme either...:).
>
> Anyway, there should be something like <scipy>/doc/ProgrammingGuidelines,
> I would suggest to start with the second attached file and put all
> "enforced" rules in there. Doesn't have to be very formal (yet), just
> put in the rules, if they are not clear that *will* be discussed on
> the list at some point, you can be sure:(

I added your comment to the following page:

http://www.scipy.org/site_content/tutorials/formatting_guidelines

> PS to Eric: Regarding patching: I think if you are on Windows you
> should save the patch as DOS-text, not Unix-text? Sorry, I do all this
> stuff within Cygwin, I really don't get any work like that done in
> Windoze.

The Mingw32 tool set does suprisingly well at bridging the gap.
Sophisticated make files don't work, but much of the other stuff does --
most of the familiar Unix commands are available.  I use cygwin some also
though, and Unix quite a lot.

see ya,
eric