@mjgardner @icing actually, being liberal in accepting things is not a good idea either. If it violates the protocol, eject, close, kill, abort. At once. That leads to better code and protocols in the longer run.
Today's my last day at Google 🥲
I'm taking some time off to focus on my family, personal health, and to work on my bots.
If anyone knows of a role that'll fit me, I'll appreciate the intro! 😊
@tewalds That would break backwards compatibility in a much more subtle way, and create a situation where its difficult to tell whether or not you are in compliance. You wouldn't be able to just grep for uses of `utcnow`, and we would have no way of warning you that at some point your code will break.
There's also no easy way to "opt-in" to the new behavior like there is with just not using `utcnow` and `utcfromtimestamp`.
It is better to remove these entirely.
It's relatively easy to make a drop-in replacement for these, but also we're deprecating them because they're conceptually the wrong thing to do, so it's best to migrate to using aware datetimes if possible: https://blog.ganssle.io/articles/2019/11/utcnow.html
`datetime.utcnow` and `datetime.utcfromtimestamp` will be deprecated in #python 3.12: https://github.com/python/cpython/issues/103857
If you maintain a package, now is probably a good time to grep your source code for `utcnow` and `utcfromtimestamp` to get out ahead of the deprecation warnings. 📅🕐
@JakubKuczys In this situation I am fine with using system pip because it's a docker container, the whole system is basically one big virtual environment.
@cnx @acdha Yes, but the way it works is that you create a dm-crypt partition seeded from `/dev/urandom`, then you mount it and fill it with 0s. The encryption translates the 0s into random bytes.
Here are the instructions I usually use: https://wiki.archlinux.org/title/Dm-crypt/Drive_preparation
@acdha Yeah, this is an empty drive that I'm about to encrypt. Filling it with random data is the first step, which I may be cargo culting, but IIRC the idea is that it prevents leaking the amount of data on the drive or something.
@acdha Just in time for the second drive to arrive so I can start on that one. 😅
@acdha Hah, good call, I had *just* remembered that I usually do... something with block sizes when I saw this.
Speed is still trending down so maybe not a steady state yet, but looks like maybe a 4-5x speedup, so this might be done in only 16 or so hours!
@adamw Definitely is part of Python: https://peps.python.org/pep-0453/
You should absolutely include your tests in your coverage measurement.
https://nedbatchelder.com/blog/202008/you_should_include_your_tests_in_coverage.html
@coveragepy Oh in the issue I guess you mention you just do line-wise regex matching.
@coveragepy How bulletproof of a solution do you want here?
Do you already have some model of the structure of the code or the context of a given method definition?
Solutions needed: Is it possible for coverage.py to automatically exclude Protocols from measurement?
Programmer working at Google. Python core developer and general FOSS contributor. I also post some parenting content.