Date: Wed, 18 Jan 2012 17:14:38 -0600 From: Mark Linimon <linimon@lonesome.com> To: CeDeROM <cederom@tlen.pl> Cc: delphij@freebsd.org, freebsd-bugs@freebsd.org, d@delphij.net Subject: Re: misc/164290: FreeBSD 9.0 IS NOT YET PRODUCTION/STABLE Message-ID: <20120118231438.GF372@lonesome.com> In-Reply-To: <CAFYkXjnnoem1kU1Ae_GHJrFW7m1mHACHa%2B4nPn=s07Tbanv=Nw@mail.gmail.com> References: <201201182118.q0ILIYaq079339@freefall.freebsd.org> <CAFYkXj=s_bOqEjxitGLz-cXMsebWaNqtRS2B2=ST4EzKk4oc3w@mail.gmail.com> <4F173ADE.5030606@delphij.net> <CAFYkXjnnoem1kU1Ae_GHJrFW7m1mHACHa%2B4nPn=s07Tbanv=Nw@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Jan 18, 2012 at 10:59:16PM +0000, CeDeROM wrote: > i always liked this logical consequence of freebsd, when something was > stable it was stable, otherwise it was devel. right now it seems to > mark devel as stable just as linux does which i dont like :-) "stable" means "we guarantee the Applications Programming Interface will remain compatible through all releases from this branch." That is all that it has ever meant. See the FAQ entry on this topic. (fwiw, this is something that Linux does not guarantee.) The only way FreeBSD can make API changes without breaking this contract is when we do a .0 release. Thus, there is pressure to do 0. releases to get changes that people are demanding. In terms of whether a .0 release is suitable for any given installation, that has to be determined by each particular user. There are a lot of people who are using what has become 9.0, including myself, under heavy load. We would not have released if it did not install, and run in a solid fashion, on most systems that the developers have access to. Since you haven't told us what type of system it is, we have no idea whether we made a good judgement call or not. fwiw, given the number of absolutely awful BIOSes out there, it will never be possible for us to run on every single machine. Check the PR database for a good survey. (Laptop BIOSes seem to be worse than desktop ones, which all seem to be worse than server ones.) We would need a lot more volunteers and direct access to all the affected machines. > also i dont like changing the good stuff only to make things "new" > just as the bsdinsytall replaced good sysinstall. sysinstall was an absolute bitch to maintain, and people have been demanding more modern features for it for years. Atttempt after attempt has been made to rewrite or update it, all of which have been abandoned up until now. Now we have something that "needs more work" but is something that we can add on to and maintain in the future. Before, we had bitrot. Sometimes you need to throw out a piece of junk and start over and that's what happened with bsdinstall. We try not to introduce regressions with .0 releases (or indeed any release), but at some point if you are going to respond for demand for new features you have to set some point as a release point. Consider "new features" and "stability with no regressions" as software goals that are inherently contradictory. If we waited for everything to work perfectly, we would never release; if we never release new features, we simply become irrelevant. I'm sure that there are many other people that are going to find that 9.0 is not suitable for their needs. This is why we have point-releases off the older branches. mcl
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20120118231438.GF372>