You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm also curious - while it's good to upgrade dependencies in general, is there something specific that's making you ask for matching versions with the official Windows releases? Should we add a test/automation/something to ensure that our version is at least as new as the official build?
I'm also curious - while it's good to upgrade dependencies in general, is there something specific that's making you ask for matching versions with the official Windows releases? Should we add a test/automation/something to ensure that our version is at least as new as the official build?
I was suggested to use prepared uv builds of python rather than official python.org.
Except for the way it's compiled, like tail-call, I would consider differences somewhat a bug.
So, I compared and found:
nice differences: python-3.12.11 compiled for windows, no \docs or news.txt
no-reason differences: sqlite not same version.
maybe some smaller differences after that, but waiting this one fixed to check more.
Except for the way it's compiled, like tail-call, I would consider differences somewhat a bug.
Just to be clear, using different versions of dependencies is not going to be considered a bug here. For other platforms, e.g., Linux, there are no official builds and, for example, we may prioritize consistency in versions across our distributions rather than match the versions used on Python.org for Windows. We also may upgrade versions before Python.org does.
While I think it's reasonable to minimize the differences, we take correctness and bugs seriously so I think it's important to clarify this point.
Remark on Windows builds: python-build 3.12.11 of sqlite3.dll is sqlite-3.47.1, while official cpython-3.12.10 uses sqlite-3.49.1 · Issue #666 · astral-sh/python-build-standalone
Activity
3.49.1.0
#667zanieb commentedon Jun 22, 2025
We can just upgrade, thanks!
stonebig commentedon Jun 28, 2025
Why does it seems not that simple ?
zanieb commentedon Jun 28, 2025
Alas they changed their build system so we actually are encountering a hiccup at #667 (comment)
geofft commentedon Jul 1, 2025
I'm also curious - while it's good to upgrade dependencies in general, is there something specific that's making you ask for matching versions with the official Windows releases? Should we add a test/automation/something to ensure that our version is at least as new as the official build?
stonebig commentedon Jul 3, 2025
I was suggested to use prepared uv builds of python rather than official python.org.
Except for the way it's compiled, like tail-call, I would consider differences somewhat a bug.
So, I compared and found:
zanieb commentedon Jul 3, 2025
Just to be clear, using different versions of dependencies is not going to be considered a bug here. For other platforms, e.g., Linux, there are no official builds and, for example, we may prioritize consistency in versions across our distributions rather than match the versions used on Python.org for Windows. We also may upgrade versions before Python.org does.
While I think it's reasonable to minimize the differences, we take correctness and bugs seriously so I think it's important to clarify this point.
Bump SQLite to `3.49.1.0` (#667)