As has been previously discussed, the project does not actually define what internal things are meant to be public apis, and what are not. So that is one more reason to consider removal of un-used functions an improvement / not for patch releases. Some third party plugin, for custom indexes, custom query handlers, etc could be using that code so removing it would break a user upgrading to a new patch version.
-Jeremiah > On Jan 25, 2022, at 7:15 AM, Joshua McKenzie <jmcken...@apache.org> wrote: > > I consider removal of dead code and refactoring to be in the same category: > critical to the longevity of the project, and not adding stability value to > GA releases which is our only priority for them. > > So improvement, and thus trunk / unreleased only. > > ~Josh > > On Mon, Jan 24, 2022 at 8:08 PM Ekaterina Dimitrova <e.dimitr...@gmail.com > <mailto:e.dimitr...@gmail.com>> wrote: > Personal opinion - I see removal of dead code as improvement which still does > not qualify for patch release. I am not afraid of breaking the tests we have > but of those we might be missing :-) > Of course, depends also on what and how much dead code we talk about > > Best regards, > Ekaterina > > On Mon, 24 Jan 2022 at 14:13, Stefan Miklosovic > <stefan.mikloso...@instaclustr.com > <mailto:stefan.mikloso...@instaclustr.com>> wrote: > Hi, > > scripts for Windows support for Cassandra were dropped in 4.0.0 (as > well as in beta4) as part of (1). The thing is that this work was not > done fully because the code itself (Java sources) were not modified > and they still contained a lot of Windows-specific logic. > > Hence, we are removing that in (2). However, and this is the core of > the issue I am getting to, we are not sure if it should be removed in > 4.0.2 too or it should be removed only in 4.1. > > The reason behind not removing it in 4.0.2 is that it might > destabilise the codebase of 4.0.x. To add to it, 4.0.x releases (or > any patch release for that matter) should only include fixes / > critical patches, not new features. The question is - is removal of > the dead code "bug" so it qualifies to be removed in a patch release > or is it a new feature which should be "introduced" in 4.1 only? > > If we remove it in 4.1 only, there will be Windows-specific code which > is not invokable, as we removed Windows startup scripts so folks > wanting to run 4.0.0 on Windows would literally have to put these > scripts back from pre-removal times. > > Regardless of this specific issue, I would like to know what the > general consensus about the removal of some dead code in patch > releases is. > > Thanks and regards > > (1) https://issues.apache.org/jira/browse/CASSANDRA-16171 > <https://issues.apache.org/jira/browse/CASSANDRA-16171> > (2) https://issues.apache.org/jira/browse/CASSANDRA-16956 > <https://issues.apache.org/jira/browse/CASSANDRA-16956>
