Black ruff mypy #7
Conversation
|
Spare me some time to fix CI. |
Sure. Suggest you go over the commit text too, you will find things that require eyes to review. From QA perspective a test that can actually test hsr, would be good. In the mean time I'll ty around a few things. |
|
@erthink mentioned on telegram. MDBX supports zero-size keys and values, which can be interpreted as NULL. That means that the signature bytes | None would be actually correct. It obviously depend if we want to expose it as b'' or None. I guess adding a test for such situation is good. |
|
Then I need to figure out in what case None is a current 'valid' reply. But I'll take a peak at that too. I'll wait for your CI setup, and review. |
This fixes the issue where MDBXCursorOp was misaligned, and failed SET_RANGE. Implements a SET_RANGE unit test.
Allow rw_traaction to pass the parentxn Implement a unit test to test a parent_txn
Implement a test that checks for null byte termination and length.
I think our users expect that when the provide bytes (or in that respect an integer converted to bytes) this is opened as-is. The option mdbx_dbi_open is effectively null terminated.
|
py3.9 will die in 1 month: https://endoflife.date/python Let's bump 3.10 |
Every function is ran against black, ruff and mypy.
I think we should still do some work to make sure that the get and iter functions will not result "None, None" depending if that is feasible from mdbx. I think it would make the code checking a lot safer, and less bytes|None, bytes|None.
There have been made some significant changes, including evident bug fixes. I would suggest, if merged, not to squash, but keep the individual commits.