On Tue, May 27, 2025 at 1:23 AM David Redondo <k...@david-redondo.de> wrote:
> Am Montag, 26. Mai 2025, 13:52 schrieb Martin Riethmayer: > > Hi all, > > [...] > > 1.) Do a "no-string-changed-bugix" release for .6 > > Pro: > > - should not have any issue with extra branching > > - consistent with released schedule (not super important, probably) > > Con: > > - Hard to enforce / communicate > > - bound to fail sooner or later > > - Problematic for CI > > I don't understand why it should be hard to enforce? > String changes are forbidden anyways in stable releases. > > What is bound too fail there? What are the problems for CI? > The problem for CI is that while the core of the system is reasonably capable of handling a multitude of branches, the seed jobs are not - and only support one latest (master) and one stable branch. This may not seem like a major issue, except that our underlying CI images may need to be rebuilt at any time. Due to the nature of us using a rolling release Linux distribution for the Linux images, BIC issues can and do occur from time to time when these rebuilds happen, which we normally resolve by running a seed job. That means that from the point the next stable release is branched (which for Plasma/6.4 I believe has already happened) until the final release of the previous stable branch (Plasma/6.3 currently) takes place you are in essence hoping that such a BIC event does not take place (because if it is does then your CI is broken) While not relevant for Plasma, this sort of situation (known as the "three branch problem") already causes a small number of issues within applications when it comes to CD builds so it would be preferable that it did not spread further. > > If there are really problems we need to look into them instead of > catch-all phrases that say nothing. > > David > Thanks, Ben
