Doesn't anything work around here?

Christian Baer christian.baer at uni-dortmund.de
Sun Nov 4 03:17:54 PST 2007


On Thu, 1 Nov 2007 18:22:25 -0500 Mark Linimon wrote:

> Sorry to come in on this discussion late.  I am behind on email.

Ain't we all? :-)

> I'm one of the people who goes through the ports and marks them
> broken -- at least on the basis of the build cluster runs.  As of
> the last complete run on sparc64-6, I think I did indeed mark those.

Were are these marks? I don't really expect them to jump right in my face
when I try to compile a port (like Seti at home did) but I'm not sure I want
to have to did for ages to find the marks either.

> The latest sparc64-7 run is continuing.  With the limited number of
> sparc64 machines we have, the elapsed time for even _incremental_
> builds is on the order of 3-5 weeks, depending on what's changed in
> the meantime.  Once that gets done, I'll probably do another pass.

That may change a little once the much faster III CPUs (or rather: the
chipsets beneath them) are supported. We won't be anywhere near the CPU
power that i386 has (the Pentium IV and AthlonXP did go pretty far there)
but at least we will be at up to 1GHz.

> As for ports that compile and install correctly, but just don't work,
> we rely on our user base to file PRs.  But there is a bit of chicken-
> and-egg problem: not many maintainers have access to these machines.
> Much more so than on i386, we are reliant on user fixes.

Keepassx probably isn't used all that much by itself and it quite possible
that only a handfull of people tried it under FreeBSD/sparc64. So I am not
really shocked that noone noticed that (yet). However, the same certainly
will not apply for Firefox and Thunderbird which are quite common apps on
Unix plattforms. So there should have been some input there.

> I personally think it's still worth putting work into sparc ports,
> if for nothing else the potential for a solid, working, FreeBSD/sun4v
                                                                  ^^^^^
Someone is working on FreeBSD/sparc32?

> down the road.  But with the resources we have right now, the priorities
> are: i386, amd64, then sparc64.

I realise this. FreeBSD started out on i386, even Alpha came a fair bit
later (btw. where is that along the list). I also realise that FreeBSD and
sparc64 is still a quite exotic combination and will probably remain that
way for some time to come. So sparc64 won't be the most important
plattform to FreeBSD. There are no complaints about this.

I'm just wondering how two really common apps that have been broken for
quite a while now could just slip through the grid.

Regards,
Chris


More information about the freebsd-sparc64 mailing list