initial work to remove py2 from the codebase

[mmerickel]

• remove references to PY2
• remove conditional imports
• figure out what to do with pyramid.compat going forward (more on this below)
• remove api docs for pyramid.compat
• remove usages of pyramid.compat.native_
• remove usages of string_types, integer_types, and other foo_types from pyramid.compat
• replace usages of pyramid.compat in codebase
• fix starter cookiecutter
• fix docs for supported versions
• fix wiki tutorial
• fix wiki2 tutorial
• fix quick tutorial
• fix quick tour docs
• fix docstrings / api docs for affected methods (track down references to bytes, unicode or python versions)

One note is that it's important to remember to keep using text_ and bytes_ where necessary for apis that may legitimately accept bytes or strings.

what to do about pyramid.compat

pyramid.compat is a public api and addons rely on it for some compatibility features.

If we want to remove it, we should push a bugfix release in pyramid that deprecates the module to help people when upgrading.

closes #2903

[digitalresistor]

If we want to remove it, we should push a bugfix release in pyramid that deprecates the module to help people when upgrading.

Yes.

[stevepiercy]

@mmerickel for docs updates, where do you want me to push? I can either create a new PR or push to your branch.

I can do similar work on the cookiecutter as well.

[mmerickel]

You can make a PR to my branch.

[mmerickel]

So at this point pyramid.compat is completely removed from this branch and the parts we want to keep are moved into pyramid.util. This is a good starting point for deciding if we want to bring anything back. I'd really appreciate some input!

[stevepiercy]

PR submitted for Python 2 removal from docs.

• The traversal wiki docs/tutorials/wiki/* needs some love. It was written for Python 2 and supposedly won't run on Python 3. I excluded that for now.
• I excluded docs/api/compat.rst because it drops some stuff and moves other bits into pyramid.util.
• There are also a couple of times where env27 and Python 2. appear in command output, but they are inconsequential.

[digitalresistor]

I asked this in private too, but native_ and text_ are not equivalent, have you verified that there was no caller depending on the behaviour of native_ to coerce to a string type?

[russellballestrini]

I almost expected there to be a lot more deletes.

[stevepiercy]

@mmerickel I noticed there is no docs/api/util.rst. Should there be one?

If so, are there any things that should not be documented?

I started on the task "fix api docs for pyramid.compat" and slammed on the brakes when I realized there was no place to move documented things from compat to util.

[stevepiercy]

I completed through the authorization step and pushed commits to the branch. I reckon y'all will need significant time to review them and walk through the tutorial.

[mmerickel]

looks like mmerickel#8 is the PR referred to above

[stevepiercy]

All changes are done for the traversal wiki in mmerickel#8 (comment)

[slav0nic]

@mmerickel why not also replace old-style formatting %s with .format, drop object for new-classes, user super() without params too?
super(QueryStringConstantCacheBuster, self).__init__(param=param) -> super().__init__(param=param)

class FileIter(object): -> class FileIter:
__nonzero__(self) -> __bool__(self)
etc

[mmerickel]

Those would be contributions we could review in another PR but weren’t the focus of this one. The string formatting is the least interesting of the ones you mentioned.

[digitalresistor]

Using .format() also has additional overhead compared to % based interpolation.