Hi, Antoine - thanks for the reply. We now understand that CI failures are common (unfortunately). We'll just need to be careful and ensure that our code does not add to the failures.
Regarding the starting from git main - we will discuss internally. We are aware of the pains of merging :) Thanks again, Marco. On Tue, Nov 25, 2025 at 11:58 AM Antoine Pitrou <[email protected]> wrote: > > Hi Marco, > > Two things: > > 1) I recommend starting from git main, because otherwise you may run > into conflicts when submitting your work to the Arrow repo > > 2) It is frequent for unrelated CI failures to appear, unfortunately, > due to issues with third-party dependencies or CI platforms > > Regards > > Antoine. > > > > Le 25/11/2025 à 18:36, Marco Arguedas a écrit : > > Hello, > > > > I am part of a team that is working on an Arrow fork, working on an > > enhancement to the C++ implementation of Parquet Modular Encryption. > > > > We have been seeing a lot of failing github actions and want to > understand > > these better. We created a fresh fork off of the most recent Arrow > release > > (22.0, commit 5eabf), and then just added a comment to each file that we > > touch on our project. No functionality was modified. The purpose of this > is > > to trigger all actions that our code changes will eventually trigger, but > > from a stable version. However, even from the stable version, we have > > noticed plenty of actions failing. > > > > This is a run of the actions: > > > https://github.com/sofia-tekdatum/arrow-22/actions/runs/19181753453/job/54839949581?pr=2 > > This is the PR (as you will see, we only added comments to a subset of > > files) https://github.com/sofia-tekdatum/arrow-22/pull/1 > > > > Most of the failing actions are Ruby-related, but there are a few other > > types in there as well. If necessary, we can characterize/summarize the > > failures in depth. > > > > Our main question here: is there a reason for a stable version with no > > virtually no changes shows a relatively high number of Github action > > failures (such as the ones above)? > > > > Regards, > > Marco Arguedas > > > >
