5 ms·
What is causing this to be a major release vs not just being a point release? Genuinely curious, since I can't gleam that from skimming the release notes.
by alberth 3y ago
What is causing this to be a major release vs not just being a point release?
Genuinely curious, since I can't gleam that from skimming the release notes.
- throw0101b 3y agoAPIs are stable with-in major version numbers. Major releases allow for API/ABI alterations. * https://docs.freebsd.org/en/books/porters-handbook/versions/ https://docs.freebsd.org/en/books/porters-handbook/versions/
- alberth 3y agoUnderstood, but I'm not seeing what API/ABI breaking changes occurred hence why I ask. https://www.freebsd.org/releases/14.0R/relnotes/ https://www.freebsd.org/releases/14.0R/relnotes/ The "Runtime Libraries and API" section doesn't indicate any breaking changes, nor am I seeing it anywhere else in the release notes.
- LastTrain 3y agoBreaking changes are allowed, not required.
- throw0101b 3y ago* 004: 14.0-CURRENT after kernel interfaces for dispatching cryptographic operations were changed. * 008: 14.0-CURRENT after changing the internal KAPI between the krpc and NFS modules. * 010: 14.0-CURRENT after changing the sndstat(4) ioctls nvlist schema and definitions. * 022: 14.0-CURRENT after commit e1a907a25cfa changed the internal KAPI between the krpc and nfsserver modules. * 026: 14.0-CURRENT after changing the internal KAPI between the nfscommon and nfsd modules. * 037: 14.0-CURRENT after removal of thread argument from VOP_STAT(9) and fo_stat. * 043: 14.0-CURRENT after removing the unused thread argument from NDINIT(9). * 054: 14.0-CURRENT after changing irq_work_queue to return a bool in LinuxKPI to match 5.10 API. * 062: 14.0-CURRENT after struct kinfo_file changes. * 086: 14.0-CURRENT after vn_lock_pair() argument changes. See: * https://docs.freebsd.org/en/books/porters-handbook/versions/ https://docs.freebsd.org/en/books/porters-handbook/versions/ Kernel APIs are also part of the stability guarantee (unlike Linux).
- drewg123 3y agoOne such changes is moving from OpenSSL 1.x to OpenSSL 3 in the base system, due to OpenSSL 1.x going EOL. That's one of the things that caused this release to take so long.
- emaste 3y agoI suspect your question is essentially "why is this 14.0, and not 13.3?" And the answer to that question is that this is a new release from our development branch, not an update to an existing branch used for the 13.x releases. FreeBSD 14.0 represents over two and a half years of feature development, stability and security improvements, and bug fixes. Some of these changes were cherry-picked into the stable/13 branch and were included in the FreeBSD 13.1 and FreeBSD 13.2 minor releases built from there. A sibling comment notes that APIs or ABIs may change between major versions. This is true, but this is not the reason for a new major version. Rather, ABI and API changes are allowed during the development cycle on the main branch, but are not cherry-picked into the existing stable branches. I have a proposed change to include in the release notes some significant changes that happened to have been merged into 13.x already, and so far have been excluded from 14.0's notes: https://reviews.freebsd.org/D42546 https://reviews.freebsd.org/D42546
- alberth 3y agoThanks so much. And really appreciate all you & the other FreeBSD developers do. Question: I’d be curious to hear your take on below (with the exception of the better defaults blog topic, since I see your comment history on that). https://news.ycombinator.com/item?id=37886507 https://news.ycombinator.com/item?id=37886507
- grahamjperrin 3y agoDeprecations such as mergemaster do not occur with point releases. There's more, but that's the first thing that comes to mind.