+1 better to deprecate and remove it than have it bit rot along. As Aldrin says, we can always revive it if the bandwidth and interest is there.
Am Mo., 5. Mai 2025 um 15:33 Uhr schrieb Aldrin <octalene....@pm.me.invalid>: > > I think deprecating is a good idea. I haven't had time to try and maintain it > and I'm doubtful the original author is following any of the communications. > > If I get around to picking up [1], then I can see about "reviving" skyhook, > but in that case the component will look very different anyways. > > [1]: https://github.com/apache/arrow/issues/40583 > > > Sent from Proton Mail for iOS > > > On Mon, May 5, 2025 at 01:13, Raúl Cumplido <rau...@apache.org> wrote: > > I am in favour of deprecating it. > > We already discussed deprecating it in February when the job was failing on > this PR comment section [1]. > > [1] https://github.com/apache/arrow/pull/45538 > > El lun, 5 may 2025 a las 10:05, Antoine Pitrou (<anto...@python.org>) > escribió: > > > > > Hello, > > > > "Skyhook" is a little-known C++ component that interfaces Arrow with the > > Ceph distributed filesystem. It received it last non-trivial change in > > 2022: > > > > https://github.com/apache/arrow/commit/546c3771a209cbcac5e03cf26e07bcd8c9601d5a > > > > You won't find much documentation for it except for an example and some > > API docs: > > https://arrow.apache.org/docs/search.html?q=skyhook > > > > Most Google search results point to the 2022 paper that presents the > > original work: > > https://www.google.com/search?q=%22arrow%22+%22skyhook%22 > > > > A C++ code search only brings Arrow C++ itself (there are unrelated C++ > > projects that use the word "skyhook"): > > https://grep.app/search?f.lang.pattern=c%2B%2B&words=true&q=skyhook > > > > Should we deprecate the Skyhook component and remove it in a release or > > two? > > > > Regards > > > > Antoine. > > > >
