@zzzeek, @maggyero, I'd appreciate your review on this, if only by building and testing it, and/or taking a look at the added tests.

@AlexWaygood would you mind taking a look at the doc changes?

For external reviewers, here's some help on how to get started reviewing this PR:

• Rationale / proposed solution
  • First, read up on the discussions in the linked issue, and those linked from that issue
  • Are there holes or inconsistencies in the proposed solution?
  • Does the proposed solution actually solve the problem?
• Implementation
  • Does the implementation align with the proposed solution?
  • Does the implementation align with PEP 249?
  • Are there logical inconsistencies in the implementation?
• Tests:
  • Do the added tests test the actual implementation?
  • Could the added tests be more specific?
  • Are there missing corner cases?
  • Are there superfluous tests?
  • How is the code coverage? Are there uncovered functions/lines/branches?
• Docs:
  • Are the added docs helpful for the average user? (Too technical? Missing examples?)
  • Are the added docs aligned with the actual implementation
  • Are there spelling/syntax errors?
  • Is the NEWS entry understandable for the average user? Does it explain the change in a concise way?
  • Can the What's New entry be improved?
• Backwards compatibility
  • Does the proposed solution follow PEP 387?
  • Could we add more emphasis on the deprecation notices?

cc. @bitdancer

Thanks, Alex! 🚀

@maggyero: I am going to make a quick review today, so can you hold the merge a little bit more?

Thanks, I definitely welcome more eyes on this. The second 3.12 alpha is scheduled for today (2022-11-14); I'm aiming for alpha 3, which is scheduled for 2022-12-05. If you can make your review by the end of November, that would be nice.

Heads-up: you should take a look at the latest emails on the db-sig mailing list, so you have an up to date understanding of how the upcoming PEP 249 autocommit attribute will look.

EDIT: I mixed up the dates; since alpha 2 is scheduled for Monday, I'm going to aim for that with this PR. If you don't have time to review this now, @maggyero, feel free to do a post-merge review. If we need to change the implementation, we can do so all the way until the feature freeze in May (IIRC).

EDIT: I mixed up the dates; since alpha 2 is scheduled for Monday, I'm going to aim for that with this PR. If you don't have time to review this now, @maggyero, feel free to do a post-merge review. If we need to change the implementation, we can do so all the way until the feature freeze in May (IIRC).

Thanks for the information and sorry for the delay, restoring the whole commit mode context took me longer than I expected. I actually did a complete rewriting of a post of mine on Stack Overflow on this topic to have a clear mental model.

Heads-up: you should take a look at the latest emails on the db-sig mailing list, so you have an up to date understanding of how the upcoming PEP 249 autocommit attribute will look.

Yes, I did.

Thanks for the information and sorry for the delay, restoring the whole commit mode context took me longer than I expected. I actually did a complete rewriting of a post of mine on Stack Overflow on this topic to have a clear mental model.

No problem. I'll merge this tomorrow evening (CET) so I'm sure we get this into the next alpha. If you manage to do a review by then; good. If not, I'd be happy with a post-merge review, if you are fine with that.

To everyone involved here and on the issue(s), thanks for fruitful discussions and helpful reviews. I'll handle the eventual deprecation of ´isolation_mode´ in a follow-up issue/PR; I'll close gh-83638 with this PR.

@maggyero, I'm landing this now. If you have further remarks, comment on the issue and we'll do a follow-up PR. Thanks for the review!

Thank you very much for your work and patience @erlend-aasland and the reviewers, I think you did an amazing job! This will improve the user experience significantly. According to the SQLite engine documentation, SQLite is the most widely deployed database engine and might even be the single most widely deployed software component of any type, so the SQLite3 driver for Python has to be of top quality.

I am fully convinced by the design (thanks @malemburg), implementation and tests. The documentation looks good to me, and we could address the remaining suggested changes in a follow-up documentation PR.