Conversation
|
@GaetanLepage Would you mind reviewing this? |
|
|
One of the python tests fail... Have you been able to build it on your end? |
|
@GaetanLepage I was able to get the Python DuckDB module's tests to pass. Just needed a couple flags changes to reflect updated flags in the upstream CI. |
GaetanLepage
left a comment
There was a problem hiding this comment.
Diff LGTM, but from my initial testing, they are a lot of packages failing. They are probably not related to this change, but it is still worth it to fix them or properly marking them as broken.
|
Details |
|
That should've been fixed with the pytest flags I added. Are you using my latest commit? |
|
Hm, I don't see my changes here, so maybe I forgot to push them! |
00b76fd to
e47ba42
Compare
|
|
|
UPDATE: The following test errors are not due to your changes. Fixed in #416520
|
|
It's time for a final bit of polish after you disable those two tests. Kindly squash and rebase your commits into a tidy sequence. It will look something like this:
|
sarahec
left a comment
There was a problem hiding this comment.
Looking good! Needs a few final changes (and I suggested a couple of optional upstream patches.)
There was a problem hiding this comment.
This would also make a lovely upstream patch.
There was a problem hiding this comment.
There's probably a reason for the constraint here, and I don't have the time right now to go digging around in the DuckDB codebase/going through the processes to figure out why.
There was a problem hiding this comment.
Turns out they had some build issue that caused them to set the upper bound: duckdb/duckdb@13db15b
There was a problem hiding this comment.
Turns out that setuptools_scm 8.0, 8.1, and 8.2 all triggered the build issue, but >= 8.3 doesn't
There was a problem hiding this comment.
Hello, my I ask a few questions for my own understanding and learning?
Was this change necessary due to failures on a specific architecture?
When I attempted this update, I only encountered problems with test/issues/general/test_17757.test, and there was no need to modify other parts of the derivation. At least on aarch64 linux and macOS.
Additionally, there is an excluded test, which is linked to an upstream issue that has now been closed, and the test passes. When should we take care of this type of cleanup? (I'm talking about test/sql/copy/file_size_bytes.test)
cfd63fb to
b26e290
Compare
|
|
As you can see, we're making progress on your nixpkgs-review. Your change uncovered a few packages that were broken and it had a cascading effect. I'm chasing down a different bug this morning and will take a closer look at your review errors this afternoon. |
|
Update: Fixed! |
|
|
`nixpkgs-review' Failures:
|
|
@GaetanLepage We have a handle on most of the remaining build breakages, do you want to go ahead and merge? |
|
It seems the aarch64-linux build failed |
@kuflierl did you see any logs you could share? Or the list of built/broken packages from |
I rather ment the of of build at https://logs.ofborg.org/?key=nixos/nixpkgs.409698&attempt_id=1d013a3f-4664-471d-aec6-7b7b00c77922 |
Thank you. That one test has given us trouble off and on. I'd disable it on Linux. Update: Yes, it's a known issue. Let's disable with a comment pointing to the issue. duckdb/duckdb#17757 (comment) |
|
|
FYI, this seems to have broken |
How come we didn't catch that in |
It did broke the package on |
@vcunat thanks for flagging that. It looks one last requested change didn't get in. Fix is in #428654. |
Things done
nix.conf? (See Nix manual)sandbox = relaxedsandbox = truenix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage./result/bin/)Add a 👍 reaction to pull requests you find important.