Define generic sys.ps{1,2,3}, for use by scripts.

[bfroehle]

Issue #123 suggests that we set sys.ps{1,2} to be consistent with python. This way scripts can tell that they are being run in an interactive session (when run using execfile or %run).

I'm not sure that I agree with this mentality, but there is likely little lost in setting the values.

Values are clearly irrelevant, but I took the constants from IPython/zmq/frontend.py.

[fperez]

I'm also not completely thrilled by this approach, but we do try to minimize inconsistencies wrt plain python, so we might as well do it.

[fperez]

I was just thinking that the [], absent actual numbers inside, look a bit nonsensical. hence my suggestion to go with bare In : and Out:. Those original constants were part of a toy prototype that @ellisonbg and I wrote back in early 2010, when we built the first proof of concept of a 2-process interactive python shell using zmq. Since we just wanted to see if it ran and the basics worked, we put a simple static prompt, but made it different from the regular python >>> one to remind us visually that it wasn't the normal shell. So there's no real importance to those old prompts, and I think we should lean more towards consistency with actual ipython use.

But this is a very minor point, of course :)

[bfroehle]

Thanks for the history. I've updated the values as you suggested.

[fperez]

Looks good, thanks! Merging now.

[fperez]

Actually, hold on: now that I think of it, should we really define sys.ps3? It's not part of the default python, and anyone actually using IPython should go through our own mechanisms. Since the purpose of declaring sys.ps{1,2} is only compatibility with plain Python, I'm not sure it makes sense to add anything beyond what Python does.

[bfroehle]

One "benefit" of sys.ps3 is a user could use it to differentiate between python and IPython...

Shrug. I'm completely indifferent on the matter.

[fperez]

OK, good point :) I'll merge now.