Be more precise about types vs. classes

[ilevkivskyi]

Also remove the mention of issubclass in the description of Union, and add a note about this for other special constructs.

[gvanrossum]

We haven't defined "compatible" previously. Maybe you can avoid it and just use "supertype"? Or briefly define it here? (The IOW doesn't really do that IMO. :-)

[gvanrossum]

Could you also review PEP 483? It's referenced from PEP 484 and while it isn't part of the standard it intends to explain the theory behind things without being so distracted by the mechanics of introducing typing into Python.

[ilevkivskyi]

I included your proposed changes. I tried to make the definition of Union as simple as possible, so that I used supertype.

It is good that you mentioned PEP 483. Actually, I already started working on extending PEP 483 along the following lines:

• explain type systems in simple but precise words
• clarify classes vs types
• explain shortly why gradual typing is a good solution for Python
• extend the discussion of generic types
• add an extensive explanation of variance of generic types

I think I could make a PR soon (probably next week).

[gvanrossum]

Thank! I've merged this downstream (in the peps hg repo) too.

If you're going to edit PEP 483, the best approach may be to copy that here first and then to build on that. I'll do that now.

[gvanrossum]

(Done, and for good measure I've added PEP 482 too.)

[ilevkivskyi]

Thanks, indeed, it is a very good idea to have all the related PEPs here.