<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html><head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<meta name="generator" content="Osso Notes">
<title></title></head>
<body>
<p>Blame is as useful as ignorance and accusation.
<br>
<br>Perhaps this is the right development model for this project.
<br>
<br>I genuinely hope that OI and Illumos succeed.
<br>
<br>With regards to sparc, as much as I like the modular Solaris kernel, Dtrace etc, it makes more sense on this occassion to investigate Gentoo as it is a free project with active sparc developers.
<br>
<br>I do have a number of spare disks and sparc hardware, so can help out if things improve on the sparc front.
<br>
<br>Thank you for your time.
<br>
<br>Regards,
<br>
<br>Peter.
<br>
<br>----- Original message -----
<br>> Hello.
<br>>
<br>> I'd like to say that upstream illumos-gate is stable and is the least
<br>> our problem.
<br>> We've seen only 3 breakages in a year:
<br>> 1) DTrace change which required additional patching for some software
<br>> (IIRC mysql, mariadb, percona and perl DTrace providers were affected)
<br>> 2) One change in ipv6 handling which broke VNC server (was fixed almost
<br>> immediately after discovering)
<br>> 3) Several changes which broke packaging (were fixed soon enough, in day
<br>> or two after discovering)
<br>> And one change which could break grub on OI was fixed during review
<br>> phase.
<br>>
<br>> There were also some changes removing software (like ntfsprogs) which
<br>> required us to add it to oi-userland.
<br>> Additionally we had to add one (!) simple patch to illumos-gate to allow
<br>> building with Perl versions > 5.10.
<br>>
<br>> So, I don't say that interaction with illumos-gate dev team is a
<br>> problem.
<br>> All the times illumos devs were extremely friendly and I think that
<br>> attempts to blame them are just FUD and provocations.
<br>>
<br>> As for real problems...
<br>> If you mention SPARC - yes, it is slowly becoming second class citizen.
<br>> But as far as I know, all SPARC-related patches coming from Igor
<br>> Kozhukhov or Gary Mils were accepted.
<br>> The reason for this state is that SPARC is some kind of sacred cow for
<br>> illumos.
<br>> Noone wants to drop SPARC support (I think most reasons for this are
<br>> religious), but only several men
<br>> really do something to keep it alive.
<br>> I'm personally one of those who don't care about SPARC, so I wouldn't
<br>> say that it's the main OI problem.
<br>>
<br>> The main problem is the lack of developers. The other are coming from
<br>> this one.
<br>> I'd say that a lot of outdated software or software which can't be
<br>> easily rebuilt is a problem.
<br>> I think that software coming from JDS and XNV consolidations which is
<br>> still out of oi-userland is a problem.
<br>> I think that binary blobs (e.g. /usr/bin/make or motif) which can't be
<br>> easily rebuilt is a problem.
<br>> I'd say that lack of software usual for FreeBSD or Linux user (e.g.,
<br>> postfix, smartmontools, hhvm on server side,
<br>> fresh python, perl, ruby versions for developers or audio and video
<br>> codecs, virtualbox, wine for desktop users)
<br>> is a problem.
<br>> ---
<br>> System Administrator of Southern Federal University Computer Center
<br>>
<br>> Peter писал 13.09.2014 11:23:
<br>> > ----- Original message -----
<br>> > > On 12 September 2014 17:28, Peter <<a href="mailto:jini@zeus.net.au">jini@zeus.net.au</a>> wrote:
<br>> > > > That's rich, a project with no stable release baseline to measure
<br>> > > > binary compatibility against won't accept patches unless ...
<br>> > >
<br>> > > Actually, our stated goal of "FCS quality all the time" means that we
<br>> > > intend to treat _every commit_ in the gate as a stable release.
<br>> > > Things are only integrated when they work, and once they're
<br>> > > integrated, documented, and people are able to use them, we intend to
<br>> > > keep supporting them compatibly.
<br>> > >
<br>> >
<br>> > Building and testing only on the developers computer before
<br>> > integration isn't sufficient to be considered stable and those who
<br>> > would contribute, don't have the resources to test widely enough to
<br>> > satisfy the stable contribution only model.
<br>> >
<br>> > Current policy also denies would be community developers the benefits
<br>> > of an experimental branch; peer review and wider testing. There may
<br>> > be exceptional developers who's code is perfect, but I am not one of
<br>> > them.
<br>> >
<br>> > For the rest of us, a second set of eyes helps to develop well
<br>> > considered api's and better quality code, and this includes a skunk
<br>> > build and wider testing.
<br>> >
<br>> > If this is the Illumos development model, the community needs to find
<br>> > a way of working with it.
<br>> >
<br>> > It sounds like Joyent has an internal experimental branch and it
<br>> > sounds like that's what OI needs in order to enable community based
<br>> > developers to participate similarly to Joyent developers.
<br>> >
<br>> > I think it's also important to have stable snapshots for users to
<br>> > report bugs against.
<br>> >
<br>> > I would be prepared to contribute via OI, if this is possible.
<br>> >
<br>> > The ARM port sounds interesting, ARM64 is a different architecture and
<br>> > doesn't share the 32 bit isa.
<br>> >
<br>> > So SPARC is presently unmaintained.
<br>> >
<br>> > While I couldn't commit to maintaining it, as it is a significant
<br>> > undertaking, I could run tests and help debugging and contributing
<br>> > some improvements.
<br>> >
<br>> > Regards,
<br>> >
<br>> > Peter.
<br>> >
<br>> >
<br>> >
<br>> > > Accidents happen; negligence cannot. We may not be perfect, but we
<br>> > > _strive_ to be, and to fix bugs (or revert integrated changes) as
<br>> > > quickly as is possible so that the software remains unbroken.
<br>> > >
<br>> > > > The logical solution is to maintain a stable kernel release
<br>> > > > version at OI, then use that as a baseline for patches and submit
<br>> > > > patches upstream to Illumos from OI rather than from individual
<br>> > > > developers.
<br>> > >
<br>> > > Working on your own distribution-specific fork of illumos-gate is a
<br>> > > reasonable idea -- it's the model we use for SmartOS at Joyent. If
<br>> > > your intent is to avoid work creeping up on you, I would recommend
<br>> > > that you merge changes into your fork early, and often. At Joyent,
<br>> > > we
<br>> > > merge into the SmartOS fork, illumos-joyent, on most business days.
<br>> > > I
<br>> > > can't recall a time when we experienced serious difficulty from
<br>> > > changes we received through these syncs. We also try to minimise
<br>> > > our diff from upstream periodically by submitting chunks of our work
<br>> > > for integration into illumos-gate.
<br>> > >
<br>> > > > That will put your patches on an equal footing as those with
<br>> > > > commercial interests.
<br>> > >
<br>> > > No, the _only_ thing that will put your patches on equal footing with
<br>> > > the commercial interests is to do the software engineering required
<br>> > > to produce quality, tested changes. There are not different sets of
<br>> > > rules for different contributors -- the same attention to quality,
<br>> > > testing, sensible architecture, etc, is expected from all
<br>> > > contributors.
<br>> > >
<br>> > > > I'd ignore the FUD about legacy platforms, projects that support
<br>> > > > multiple architectures are more stable for it.
<br>> > >
<br>> > > I whole-heartedly agree that operating systems that run on more than
<br>> > > one architecture are less likely to see particular classes of bugs
<br>> > > that are hidden by some architecture-specific implementation detail;
<br>> > > e.g. byte ordering, different core system hardware drivers, etc.
<br>> > > Unfortunately, if there are no _active_ OS engineers using and
<br>> > > working on a particular architecture, it will absolutely rot over
<br>> > > time; eventually becoming an obstruction to important changes where
<br>> > > those changes require architecture-specific work.
<br>> > >
<br>> > > > With regard to sparc, there are a number of current chip
<br>> > > > manufacturers, Aeroflex, Fujitsu, Oracle, FeiTeng. How hard would
<br>> > > > it be to support sparc v8? What about ARM64?
<br>> > >
<br>> > > Robert Mustacchi, myself, and a few others, have been investigating
<br>> > > (and in Robert's case, working on) an ARM port of the OS. It's a
<br>> > > lot of work -- work which is entirely uninteresting to our employer,
<br>> > > so it's a spare time project. I would love to see it completed,
<br>> > > and hopefully we will get there in time.
<br>> > >
<br>> > > > It's clear that Illumos isn't the right place for me to contribute
<br>> > > > directly, but the whole community will benefit if I and others
<br>> > > > like me can contribute via OI.
<br>> > >
<br>> > > Why is that? If you want to procure and operate SPARC hardware, to
<br>> > > fix build issues as they arise, and to write SPARC-specific code for
<br>> > > changes that require it, why not do it in illumos? If you (and
<br>> > > others
<br>> > > like you) do not step up to maintain the SPARC bits, then they truly
<br>> > > are dead weight and need to be removed.
<br>> > >
<br>> > > --
<br>> > > Joshua M. Clulow
<br>> > > UNIX Admin/Developer
<br>> > > <a href="http://blog.sysmgr.org">http://blog.sysmgr.org</a>
<br>> >
<br>> >
<br>> > _______________________________________________
<br>> > oi-dev mailing list
<br>> > <a href="mailto:oi-dev@openindiana.org">oi-dev@openindiana.org</a>
<br>> > <a href="http://openindiana.org/mailman/listinfo/oi-dev">http://openindiana.org/mailman/listinfo/oi-dev</a>
<br>>
<br>> _______________________________________________
<br>> oi-dev mailing list
<br>> <a href="mailto:oi-dev@openindiana.org">oi-dev@openindiana.org</a>
<br>> <a href="http://openindiana.org/mailman/listinfo/oi-dev">http://openindiana.org/mailman/listinfo/oi-dev</a>
<br><br></p>
</body>
</html>