Conditionally patch time.clock

[pganssle]

time.clock was removed in Python 3.8, which is causing freezegun to fail when importing the API module as it tries to stash and patch the clock function. This makes the patching logic conditional on the existence of a time.clock function.

time.clock has been deprecated for some time, and I guess it was just missed in at least my test suite because the real function was never invoked.

It may be worth hard failing on deprecation warnings in your test suite in the future, to avoid problems like this.

[coveralls]

Coverage increased (+0.04%) to 93.786% when pulling d10c1aa on pganssle:fix_38 into c2d1dd2 on spulec:master.

[coveralls]

Coverage increased (+0.04%) to 93.786% when pulling d10c1aa on pganssle:fix_38 into c2d1dd2 on spulec:master.

[coveralls]

Coverage increased (+0.04%) to 93.786% when pulling d10c1aa on pganssle:fix_38 into c2d1dd2 on spulec:master.

[coveralls]

Coverage increased (+0.04%) to 93.786% when pulling 35713b8 on pganssle:fix_38 into c2d1dd2 on spulec:master.

[pganssle]

Test failure on 3.8-dev seems unrelated, but also is emitting deprecation warnings in time.clock, so evidently whatever version Travis has doesn't have the time.clock removal yet.

[boxed]

Nice! Thanks.

[pganssle]

@boxed Since this is actively breaking on Python 3.8 on import, I think a release sooner rather than later would be a good idea - considering this is a testing library and it may block people from testing 3.8 before the release.

Not sure if you use the publish target in the MakeFile, but if you do I recommend instead using something like #297 for the publishing instead.

[boxed]

Agreed. But I don't have access to make a release. @spulec

[spulec]

Thanks!

Released with 0.3.12.

[pganssle]

@spulec It seems like you used the publish Makefile target without merging #297, because once again you have hit the problem from #268. Given that the release was 14 minutes ago, can you please delete it and cut a new one?

Once PEP 592 is implemented, you will be able to "yank" the last two releases and cut new releases in their place, but you may as well delete the most recent one before someone gets hit by it.

[spulec]

Sorry about that. I've merged #296 for the future.