bpo-43094: Update sqlite3 docs to match implementation

[erlend-aasland]

https://bugs.python.org/issue43094

[erlend-aasland]

@berkerpeksag Would you mind reviewing this?

skip news, I presume.

[Fidget-Spinner]

I'm not Berker, but this LGTM :).

For convenience:
create_function:

cpython/Modules/_sqlite/clinic/connection.c.h

Line 97 in 3ccef1c

"create_function($self, /, name, narg, func, *, deterministic=False)\n"

create_aggregate:

cpython/Modules/_sqlite/clinic/connection.c.h

Line 160 in 3ccef1c

"create_aggregate($self, /, name, n_arg, aggregate_class)\n"

Which seems to match up with the changes in the docs.

[terryjreedy]

This is a substantive change, not a typo or grammar fix, so I would not merge without a blurb.

[Fidget-Spinner]

Sorry if this isn't related, but I just noticed the docstrings on some other functions don't match either:
Eg create_collation docstring:

cpython/Modules/_sqlite/clinic/connection.c.h

Line 623 in 3ccef1c

"create_collation($self, name, callback, /)\n"

Ref:

cpython/Modules/_sqlite/clinic/connection.c.h

Line 633 in 3ccef1c

PyObject *name, PyObject *callable);

Meanwhile the one on python docs is correct
https://docs.python.org/3/library/sqlite3.html#sqlite3.Connection.create_collation

I wonder if this is intentional?

[erlend-aasland]

Sorry if this isn't related, but I just noticed the docstrings on some other functions don't match either:
Eg create_collation docstring:
[...]

That seems to be a regression from #23341. Good catch! I guess it's fine to include that one in this PR. I'll do an audit of all docstrings/docs/specs in sqlite3 and add any discrepancies to the PR.

[erlend-aasland]

This is a substantive change, not a typo or grammar fix, so I would not merge without a blurb.

Noted. Thanks!

[erlend-aasland]

@Fidget-Spinner: I've done an audit of the module level functions. Should this be backported to 3.9 and 3.8? If so, I will have to split this PR in two: one that contains docstring fixes (regressions from my recent AC change), and one that contains documentation fixes that should be backported.

[Fidget-Spinner]

@Fidget-Spinner: I've done an audit of the module level functions. Should this be backported to 3.9 and 3.8? If so, I will have to split this PR in two: one that contains docstring fixes (regressions from my recent AC change), and one that contains documentation fixes that should be backported.

Thanks for fixing them! My hunch is the regression fixes should go into another PR with the bpo number same as the PR which caused the regression, while the backportable stuff remains in this PR. However I am not an sqllite3 expert by any means, so I wouldn't be surprised if this hunch is wrong.

[erlend-aasland]

Thanks for fixing them! My hunch is the regression fixes should go into another PR with the bpo number same as the PR which caused the regression, while the backportable stuff remains in this PR.

Makes sense!

[terryjreedy]

Do not merge yet because there is still discussion on pydev, and some disagreement, as to precisely the fix should be. In particular, can we make the proper names keyword only? Please make sure you read the thread and wait for consensus.

Regression fixes do not have to be attached to the regressing issue. It has been done both ways, with a note on the original issue if done on a separate issue.

[erlend-aasland]

Do not merge yet because there is still discussion on pydev, and some disagreement, as to precisely the fix should be. In particular, can we make the proper names keyword only? Please make sure you read the thread and wait for consensus.

Regression fixes do not have to be attached to the regressing issue. It has been done both ways, with a note on the original issue if done on a separate issue.

Yes, I'm following the thread. I think I'll close this PR bco. noise, and open a new one when there's concensus.