h-vetinari commented on PR #42114:
URL: https://github.com/apache/arrow/pull/42114#issuecomment-2161882793
So in general, what you'd need to do is:
* sync the recipe contents (`meta.yaml` plus all relevant build & test
scripts), minus the patches and keep pointing to the local sources
* https://github.com/conda-forge/arrow-cpp-feedstock
* https://github.com/conda-forge/pyarrow-feedstock (new!)
* https://github.com/conda-forge/r-arrow-feedstock (currently broken on
windows, unfortunately)
* sync the respective variant configs
* per feedstock; e.g. from
[here](https://github.com/conda-forge/arrow-cpp-feedstock/tree/main/.ci_support)
for arrow-cpp
* ensure that the file names in `tasks.yaml` are updated where necessary
* (potentially) sync changes from [build
scripts](https://github.com/conda-forge/arrow-cpp-feedstock/tree/main/.scripts)
([example](https://github.com/apache/arrow/pull/15014))
* again per feedstock
* this is the trickiest part, because there are local changes to make
arrow CI work that AFAIU aren't documented anywhere; it basically needs careful
study of the diff, and deciding which changes to take or not. Fortunately, this
step is rarely if ever necessary.
In the case of this update, the
[split](https://github.com/conda-forge/arrow-cpp-feedstock/issues/1381) of the
python outputs into a separate feedstock means you'll have to integrate a full
third recipe here, and build the `pyarrow` bits on top of `libarrow` (like
currently done for `r-arrow` already). On the plus side, this will
automatically resolve #15045.
Some more inspiration: #35637 #34652 #14832 #14102
--
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]
