Date: Thu, 11 Feb 2021 00:07:03 -0800 From: Mark Millard <marklmi@yahoo.com> To: Mark Linimon <linimon@lonesome.com> Cc: "ppc@freebsd.org" <ppc@FreeBSD.org>, "pkubaj@freebsd.org" <pkubaj@FreeBSD.org>, "linimon@freebsd.org" <linimon@FreeBSD.org> Subject: Re: how many users of FreeBSD 12 do we have on powerpc? Message-ID: <C608E83F-CDFA-421A-81EF-0083EDA6184D@yahoo.com> In-Reply-To: <20210210223920.GA31353@lonesome.com> References: <20210210223920.GA31353@lonesome.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 2021-Feb-10, at 14:39, Mark Linimon <linimon at lonesome.com> wrote: > So I want to get an idea if we can start trying to move FreeBSD/powerpc* > users from 12.X to 13.0 once 13.0 is released. > > The problem is that we have limitations on: > > - hardware resources w/rt building ports on 3 branches Only matters for powerpc64? No such building for 32-bit is normally done? > - human resources trying to support 3 branches This still applies for 32-bit powerpc variants, outside of just ports. Although, trying to cover 32-bit PowerMacs may have a similar status for this human resources point outside of ports. (May be only chroot style or analogous on powerpc64's that allow such for 32-bit code? As stands, I do modern FreeBSD 32-bit PowerMac activity on powerpc64 via chroot and poudriere activity, other than evidence gathering relative to known 32-bit kernel problems for PowerMacs.) > The powerpc ports team (the entire handful of us) is talking about > dropping ports building for 12-STABLE in favor of starting to build > 14-CURRENT. > > How many people will this affect if we do so? > > fwiw, I personally believe 13.0 will be a complete superset except for: > > - ports that fail due to duplicate_symbol > > and all of those are problems shared with even amd64. > > (Also, I personally believe that it is fair to say "we do not really > support powerpc* on 10.X or 11.Y. Please upgrade.") > > How many people will these changes effect? It's time to speak up ... It can be hard to tell about the contrasting case(s) vs. just non-responders without explicit reports, so . . . I have converted the old PowerMac contexts that I have access to to be based on FreeBSD:14:* ABIs, including having built various ports for powerpc64 and for 32-bit powerpc (nearly the same ports for both these days). Old PowerMacs are the only form of powerpc* families that I have access to. Side note: With the 2-socket/2-core-each G5 failures I've gone from a set of origins that resulted in about 476 port builds to a set that results in about 101 port builds. No more llvm* builds, for example. I cross build FreeBSD itself for powerpc64 and 32-bit powerpc. I build 32-bit powerpc ports on the one well-operating G5 that I still have access to (2 sockets/1-core-each). I have fairly strong time preferences. But going in the other direction: It is not clear that I'd use: http://pkg.freebsd.org/FreeBSD%3A14%3Apowerpc64 to get installations of any ports that I choose not to build. Port building is a test of if the system is still operating fairly well and my historical toolchain explorations no longer serve much purpose. And there is no: http://pkg.freebsd.org/FreeBSD%3A14%3Apowerpc === Mark Millard marklmi at yahoo.com ( dsl-only.net went away in early 2018-Mar)
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?C608E83F-CDFA-421A-81EF-0083EDA6184D>