Pyfakefs usage scenarios

[mrbean-bremen]

This is more of a discussion point than an issue, and also for a later release.

I wonder if there is really the need to use pyfakefs via mock (the third scenario), given that you always can use Patcher directly if not using unittest or pytest. From what I have seen from contributors so far, none of them seems to be using the mock scenario.

Anyway, I doubt the usefulness of mocking glob, tempfile or shutil separately, but maybe I'm wrong here. On the other hand, if removing that scenario for these modules, fake_filesystem_glob and probably fake_tempfile (which is implemented incompletely anyway) can be removed, and most from fake_filesystem_shutil implementation also (though this needs some additions in FakeOsModule).

Edit: I moved some of the points into separate issues that can be implemented independently (#191, #192, #193 and #194).

What remains for this issue:

• adapt the documentation (replace mock scenario with Patcher scenario)
• remove fake glob and tempfile implementation

[jmcgeheeiv]

Thank you for thinking so deeply about this.

I generally understand and agree. Could I ask you to separate these into two or three paragraphs or perhaps even separate issues? Also, please specify "one" in "...none seems to be using that one."

[mrbean-bremen]

Okay, I created a few more issues that can be handled separately.

Do you think this is something for the next release? And thanks for the 3.2 release, by the way - you have done a lot of documentation cleanup, I liked that.

[jmcgeheeiv]

Sure, this is an excellent target for 3.3.

[mrbean-bremen]

Great!

Note I won't have access to a computer for a couple of weeks - I will finish the nanosecond time related issues afterwards.

[jmcgeheeiv]

Bon voyage.

[mrbean-bremen]

Thanks, having that!

[mrbean-bremen]

Remaining task: adapt README.MD.

[mrbean-bremen]

All done - closing.
@jmcgeheeiv - when you will be available again, you may prepare a 3.3 release. The bugs coming from TSTL are ongoing stuff, but nothing really important for the release, I think. I think the next major change would be to pep8tify the API.