Re maintaining a distro: It’s just the upcoming freebsd 13 that won’t work, freebsd 12 is/will be supported for a couple of years still.
Two people I know who did package buiding and distro maintaining (and no longer do it, stopped a few years ago). near the end were putting in half a day a week and a few hundred bucks a month for power/network costs just to keep things rolling/status quo. They’re both the kind of people who could work for anyone they want and get $250k/year easily without having to deal with customers or be managers. In both cases, hardware was mostly donated by the communities, they were your typically 2 socket server machines of the day (1-3 year old). Only special thing at the time was that storage was all flash, because you need low latencies to keep the CPUs fed while installing dependencies/building/ … Probably less than <$10k per server in hardware + time investment to keep it running, was able to keep 90% of the software fresh within 7days. (Again this was a couple of years ago, there were other people dealing with most of community stuff).
I’d wholeheartedly recommended you not to get into being a distro maintainer.
Software is slower and slower and bigger and bigger, both to build and to run, and hardware is not catching up as quickly. Most big distros are backed by medium size companies who give them lots of stuff for free and get to write it off as a business expense cause they might run that distro.
This guy has some really good ideas on how to make this whole distro business easier: https://michael.stapelberg.ch/posts/2019-08-17-introducing-distri/ - however the approach of putting systems together like that is radically different and non-traditional and scary because changes are scary and polarizing. However, it’s probably the only hope independent distro maintainers have these days.
While you can theoretically fork FreeBSD and maybe try to go through the motions and pretend you’re running something for a bit, I think it’s highly likely you’ll just fail (let alone you’ll be falling further and further behind on features). There’s better things to do with your life in 2020 than try to maintain and start a distro fork with no backing.
The other solution is to just build from source, which is why I suspect Gentoo, and other similarly “build from source” distros are likely to be the last holdouts for dying architectures (though SPARC isn’t exactly dead, as far as I can tell, but “mainstream” use of it does appear to be dying with both Oracle as well as Fujitsu pulling out of new development. The former because they are allergic to nice things, the latter supposedly because they don’t want to maintain Solaris)
I swear, Oracle buying Sun is probably the worst thing to have happened to IT in the past, errr, forever? I honestly can’t think of anything worse off the top of my head. Suggestions welcome though.
Of course, but given a less popular (or plain dead) platform I’ll take a small tested core and having to dip into “testing” or even plain “untested on this platform” recipes available through the package manager over having to jump through the hoops generally required to get something that’s not even in the tree installed on most binary distributions.
I don’t think they try to build everything, maybe everything that’s marked for that arch which a lot of stuff isn’t for the less common architectures until someone requests it.
But then again, I haven’t been keeping up with Gentoo politics and policies since around the time Daniel Robbins went to work for Microsoft. Things got a lot more complicated (and political) after that.
At one point they were trying to build everything, in a sense that they had blacklists of broken versions they were iterating on trying to forever shrink them. (I don’t know if they still do).
The call had been out for about a decade. People have said they would do things but never followed through. GCC 4.2.1 has been long overdue for removal, and from my understanding nobody has been willing to bring sparc64 up in a modern toolchain, not just in FreeBSD but anywhere.
If they drop support for Sparc64 (or other obscure/ancient platform X), that doesn’t mean the existing running versions go away. It’s not like sparc is getting new hardware to support.
You can still use it for stuff, it’s just that “stuff” would/should not include exposure to hostile environments.
Not entirely sure how this is related to SPARC, something FreeBSD specific?
New GCC’s build for sparc just fine. In fact GCC 9.2 is even stable on Gentoo for sparc as per this resource (not sure what I’m running can check that tomor…later today).
Sorry, the list archive didn’t follow the thread as completely as it appears in my email. There was a follow-up in January that went into more detail about sparc64:
I was mistaken about there not being modern toolchains for sparc64 outside of FreeBSD. The main issue was indeed that the C runtime library for FreeBSD specifically does not build for sparc64 with a modern toolchain, and nobody stepped up to fix it.
In short, FreeBSD could not be built for sparc64 anymore. There was plenty of opportunity for someone with an interest in sparc64 hardware to make it work, but nobody did. So the broken sparc64 platform has been removed. As @thro said, it’s still there in the existing releases, but it will not be present in new versions of FreeBSD moving forward.
Gentoo on SPARC? Yes. Had to borrow a built kernel from a Gentoo SPARC dev though, since the one from the install disk gave me trouble, leading to being unable to build an initramfs (permahang on the sleep config test).
That allowed me to bootstrap my own kernel and stuff’s running fine-ish now (console doesn’t come up, but it boots just fine and I can just ssh into it, so something to figure out later)
% uname -a
Linux luxuria 4.19.97-gentoo-sparc64 #1 SMP Tue Jan 28 13:02:12 CET 2020 sparc64 sun4v UltraSparc T2 (Niagara2) GNU/Linux
For completeness:
% gcc --version
gcc (Gentoo 9.2.0-r2 p3) 9.2.0
Copyright (C) 2019 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.