Conversation
|
Hi - I restarted the Windows run now since it seems spurious. But if it fails again it might need more thorough investigation. 👍 |
|
This still appears to be failing but I'm a little at a loss as to why, the same build worked fine in our extension repo: https://github.com/northpolesec/duckdb-gcs/actions/runs/22139297199/job/64015425496. A cursory search suggests the error is related to input paths being too long. |
|
I don't know if this helps but I looked through the build logs and there is a slight version mismatch, this failing PR says: but the working PR from the extension repo says: on the build step. |
|
There was some instability on CI, that I think it's fixed in main. I took the liberty of rebasing the PR onto duckdb/community-extensions main branch, let's see what CI says. |
|
Thanks @carlopi, could you try using the latest commit 9d2cb749affdebef8f4c0c78e21683bc9f5e2c0f with the updated submodule (1.4.4). I hope this resolves ix the build issue 🙏 |
|
The build has used the latest version (1.5.0)...According to the build script, to use the v1.4 |
|
🤞 |
|
The upgrade to DuckDB v1.5.0 has been done and the build was successfull for the on both my fork branch and on the main branch of the extension repository 🤔 |
|
Restarted Windows CI, let's see. |
|
Thanks, @carlopi but still fail ... It's possible to get the following log file from the build : |
|
After reviewing logs, It seems that in the community-extensions build an external ninja version |
|
I am trying to debug the issue, opened a PR at #1536 to double check the results. |
Also: re-enable Windows builds