Hi Raúl, Antoine,

Thanks — Raúl's proposal sounds like the right call to me. I agree with
Antoine that rushing the review is the wrong trade-off.
Let's proceed with RC1 as planned

A 25.0.1 patch release with GH-50327 (and GH-50430 if ready) works well for
the Spark timeline: what matters for Spark 4.2 users is that a release
with the fix exists to upgrade to, and Spark can point to it in the
migration notes. I'll keep working through the reviews at the normal pace.

Raúl — once the PRs merge, a rough sense of the patch-release timing would
help us coordinate the Spark-side documentation. Thanks for offering
this.

Best,
Liang-Chi


On Wed, Jul 8, 2026 at 11:50 PM Antoine Pitrou <[email protected]> wrote:

>
> Hi,
>
> Le 09/07/2026 à 00:31, L. C. Hsieh a écrit :
> >
> > I'd like to ask whether the community would consider cancelling RC1 and
> > cutting an RC2 that includes GH-50326 / PR #50327 (making Array.to_pylist
> > convert without per-element Scalars; reworked per Antoine's review and
> > awaiting another pass).
>
> I understand the concern, and I agree this is a long-standing
> performance wart in PyArrow, but adding a last-minute non-trivial
> improvement (for which we will have to rush review) is the best way to
> introduce regressions.
>
> The alternative would be to start the feature freeze over after
> integrating that PR. I don't think that's a tremendous idea (that would
> push back our quarterly release by ~one month), but I don't have a
> strong opinion either.
>
> Regards
>
> Antoine.
>
>

Reply via email to