How do the other languages in apache/arrow-adbc consume bits from apache/arrow before they are officially released in a major version? For example, I see Java uses the 11.0.0 version (https://github.com/apache/arrow-adbc/blob/main/java/pom.xml#L31). What would happen if it is dependent on a bug fix/feature in apache/arrow? Would the Java ADBC code needs to wait for an official release before it can use it?
-----Original Message----- From: David Coe <david....@microsoft.com.INVALID> Sent: Tuesday, May 23, 2023 10:02 AM To: dev@arrow.apache.org Subject: [EXTERNAL] ADBC -> Arrow references for C# [You don't often get email from david....@microsoft.com.invalid. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ] We recently put up feat(csharp): adding C# functionality by davidhcoe * Pull Request #697 * apache/arrow-adbc (github.com)<https://github.com/apache/arrow-adbc/pull/697>. This PR introduces C# functionality for ADBC and is dependent on capabilities introduced in GH-33856: [C#] Implement C Data Interface for C# by CurtHagenlocher * Pull Request #35496 * apache/arrow (github.com)<https://github.com/apache/arrow/pull/35496>, which has now been merged to main. We currently use the submodule approach for ADBC to reference Arrow. This submodule was pointed at [submodule "arrow"] path = arrow url = https://github.com/CurtHagenlocher/arrow branch = CSharp_CAPI until the PR was merged. Since this has now landed, we can clean up the references a bit, but wanted to get some thoughts on the best way to do so. Here are some proposed options, in order of preference and ease of use: 1. An Apache.Arrow 13.0.0-alpha release is published to nuget 2. Release an interim nuget package called something like Temp.Apache.Arrow.CData so it can be referenced from within the ADBC project 3. Update the submodule to point to the main repo. Add instructions for how to pull this submodule. Any thoughts on the best way to do these and potential timeframes? * David
