thisisnic commented on issue #50765:
URL: https://github.com/apache/arrow/issues/50765#issuecomment-5168132173
> Investigation of the CI artifacts from the failing `r-recheck-most`
Crossbow run (`#30600965885`) shows that this failure was caused by a transient
compilation error in the baseline `oldpkg` environment rather than a regression
in Apache Arrow.
> ### Log Findings
>
> 1. **`duckdb` (`oldpkg`)**:
> `duckdb` source compilation failed during the baseline `oldpkg`
(`arrow 25.0.0`) check due to a compiler process termination:
> ```
> make: *** wait: No child processes. Stop.
> make: *** Waiting for unfinished jobs....
> make: *** wait: No child processes. Stop.
> ERROR: compilation failed for package 'duckdb'
> * removing '/__w/crossbow/crossbow/oldpkg/duckdb.Rcheck/duckdb'
> ```
>
> 2. **`duckdb` (`newpkg`)**:
> `duckdb` compiled and installed cleanly against development `arrow`
(`25.0.0.9000`):
> ```
> * checking whether package 'duckdb' can be installed ... OK
> * checking tests ... OK
> Status: 1 NOTE
> ```
>
> 3. **`framework` / `pins` / `vetiver`**:
> All three packages have identical test failure counts in both
`oldpkg` and `newpkg`:
>
> * `framework`: `FAIL 3 | WARN 31 | SKIP 97 | PASS 634` (Identical
in oldpkg & newpkg)
> * `pins`: `FAIL 1 | SKIP 41 | PASS 130` (Identical in oldpkg &
newpkg)
> * `vetiver`: `FAIL 2 | WARN 2 | SKIP 71 | PASS 226` (Identical in
oldpkg & newpkg)
>
>
> Across all 167 reverse dependencies checked, **0 packages experienced
regressions** in `newpkg`.
> ### Root Cause & CI Mechanism
>
> The CI failure in the `Show results` step occurs because `r-devel/recheck`
invokes R's `tools::check_packages_in_dir_changes("newpkg-checklogs",
"oldpkg-checklogs")`. This function flags any package status difference between
`oldpkg` and `newpkg` as a change, even when a package transitions from `ERROR`
in `oldpkg` to `OK` in `newpkg`, leading to `stop("Found potential problems")`.
> ### Recommendation
>
> No code or CI changes are needed in Apache Arrow. A re-run of
`r-recheck-most` or closing this issue as transient is recommended.
Yeah, don't do this, it's not helpful and just creates extra noise. If you
want to contribute, you could look at this as a human and figure the actual
cause and a fix, but pasting AI output is the opposite of contributing.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
