Alyssa Coghlan

CPython core developer, device security architect @(watch this space), cognitive science dabbler, secular humanist, charitably mercenary cynical idealist :)

[Alyssa Coghlan @ncoghlan_dev]

Things that need trolling: authoritarian governments, corrupt multinationals, abusive "leaders". Who do weak sycophants harass? Vocal women.

[Alyssa Coghlan @ncoghlan_dev]

@dstufft Even figuring out a suitable *review* process for those early packaging ones was quite an extended saga!

[Alyssa Coghlan @ncoghlan_dev]

For anyone curious, the addition is a new "See also" block for concurrent.futures at the start of the threading docs (and a similar addition in multiprocessing): https://t.co/pyEP7eKFRP

[Alyssa Coghlan @ncoghlan_dev]

The good kind of conference driven development: identified a documentation hole based on a talk, posted a PR that evening to close it. #kiwipycon

[Alyssa Coghlan @ncoghlan_dev]

@pfctdayelise Context was that different situations will have different hierarchies. The bottom tier in this particular example was also for truly one-off code (where an Undo button is likely enough history)

[Alyssa Coghlan @ncoghlan_dev]

The scientific hierarchy of coding needs described by @Sydonahi is honestly pretty similar to what commercial software needs (just replacing scientific impact with business/bottom line impact in guiding the investment of time) #kiwipycon

[Alyssa Coghlan @ncoghlan_dev]

@s_lott Android recommended enough articles from them to realise they're throwing up clickbait to get outraged views (sometimes with articles on the same day arguing opposite perspectives without mentioning each other)

[Alyssa Coghlan @ncoghlan_dev]

High praise indeed from Tom Clark at #kiwipycon: "Context managers are no more complicated than they need to be" :)

[Alyssa Coghlan @ncoghlan_dev]

@anthonypjshaw @pyblogsal There's also a home for any CPython crashers you find that are deemed "Won't fix": https://t.co/sIrew8eSvN :)

[Alyssa Coghlan @ncoghlan_dev]

@HeyChelseaTroy @raymondh @llanga Open source living up to its goals :)

[Alyssa Coghlan @ncoghlan_dev]

@nhumrich For the benefit of anyone else reading the Twitter thread: we chatted via DM, and rather than submit a new PEP, @nhumrich is going to look into brushing the dust and cobwebs off the existing PEP 501, and becoming a co-author on that.

[@[email protected] @nickyringland]

Teachers! Just a few more days to get #pyconline student showcase applications in for @PyConAU!

Have your students made something cool using python? Maybe an assignment, side project, or game? We want to hear about it!

https://t.co/EhvONnCbq7

[Alyssa Coghlan @ncoghlan_dev]

@KathyReid @candeira (I write this as someone that let his passport expire for the first time in 20+ years a few months back, since you're still eligible for the fast path renewal process up to 3 years after expiry)

[Alyssa Coghlan @ncoghlan_dev]

@KathyReid @candeira I'm picturing an Australian Passport Office group chat:
Bob: Should we turn off the renewal reminder scheduled job? It's not like anyone can travel anyway.
Tina: Won't that create a huge renewal backlog when the borders open again?
Bob: Oh yeah, good point. I'll leave it alone.

[Alyssa Coghlan @ncoghlan_dev]

@nhumrich But yeah, even I never really warmed to the "i" prefix (over the original "t" prefix), and if we went back to proposing the "t" prefix we could at least borrow half of the JS terminology (template literals), even if "template strings" needed to be avoided due to "string.Template"

[Alyssa Coghlan @ncoghlan_dev]

@nhumrich Folks also objected to calling them strings when they may never actually be rendered to a string at any point (unlike f-strings, which are eagerly rendered by the compiler), but it was the "string.Template already exists" argument that I personally found compelling.

[Alyssa Coghlan @ncoghlan_dev]

@di_codes @MylesBorins But IIRC, the original motivation for allowing them in the PEP was just "PyPI already has projects with 4+ numeric segments in their version numbers, they're not ambiguous in any way, so there is no reason to disallow them".

[Alyssa Coghlan @ncoghlan_dev]

@di_codes @MylesBorins The specific thing I personally use a 4th digit for is CalVer-with-patch-releases. Normal versions are YY.M.serial (to allow for multiple feature releases in the same month), if a patch series is needed for some reason, it becomes, YY.M.serial.patch.

[Alyssa Coghlan @ncoghlan_dev]

With the inclusion of async generator based features, the minimum supported version is now Python 3.6+, so it will be interesting to find out if anyone is still using old Python versions with old versions of installation tools that ignore the Required-Python metadata field.

[Alyssa Coghlan @ncoghlan_dev]

contextlib2 21.6.0 has been released, updating the API to match Python 3.10 (including all the previously omitted asynchronous features from 3.7+): https://t.co/pPO03LnX8x

[Alyssa Coghlan @ncoghlan_dev]

Release notes at https://t.co/RIsxcdNzAs