Slashdot Mirror
OpenIndiana Hipster 2015.10: Keeping an Open-Source Solaris Going
An anonymous reader writes: It's been five years since Oracle killed off OpenSolaris while the community of developers are letting it live on with the new OpenIndiana "Hipster" 15.10 release. OpenIndiana 15.10 improves its Python-based text installer as it looks to drop its GUI installer, switches out the Oracle JDK/JRE for OpenJDK, and updates its vast package set. However, there are still a number of outdated packages on the system like Firefox 24 and X.Org Server 1.14 while the default office suite is a broken OpenOffice build, due to various obstacles in maintaining open-source software support for Solaris while being challenged by limited contributors. Download links are available via the OpenIndiana.org release notes. There's also a page for getting involved if wishing to improve the state of open-source Solaris.
Let it go. Your OS is dead. Let it be dead in peace.
Nobody knows it, but god damn do the linux fan-boys hate it!
How do you "kill off" an open source project if the public is willing to take over the development and maintenance? Sure you may be continuing with a non-open-source branch of the code for your own products, but that doesn't stop anyone from working with the last released code base.
I do not fail; I succeed at finding out what does not work.
Never looked much at Solaris but this thing is both a systemd-free zone AND has GNOME 2. Not bad, might take a look at it.
It still perplexes me why do they insist on keeping the desktop aspect of OpenIndiana alive instead of focusing on the server aspect where it can shine the most?
I know that Solaris did have a workstation presence at one point; back when each vendor with a pet Unix had a line of workstations to run it, usually on their pet CPU architecture; but it seems very, very, strange that they'd be focusing on desktop features at all(especially if they don't have the resources to do them properly; especially with web browsers outdated and/or broken is worse than nothing).
You certainly hear about cool stuff that Solaris has; and others either lack or have only by virtue of pulling from Solaris(Dtrace, Solaris Containers, ZFS, probably some others); but 'desktop experience' sure isn't one of them. Especially when 'the desktop' also tends to imply needing workable support for a variety of desktops and laptops of various degrees of unfriendliness, it seems a strange place to put any resources.
1. To see the date of http://hg.openjdk.java.net/jdk...
2. hg clone http://hg.openjdk.java.net/jdk...
3. cd *jdk7u* && sh get_source.sh
It requires installed Ant ALSA CUPS and FreeType.
4. It can be auto-compiled using the defunct JDK 7 or the alive OpenJDK7, it does not require JDK 6 for bootrapping, and should be specified ALT_BOOTDIR=path to the jdk.
The vitality of an open source project isn't defined by its developers. It's defined by its users!
Look at Firefox starting with version 4, or GNOME 3. They have a number of developers, but users just don't want to use Firefox or GNOME 3. GNOME 3 hasn't seen much use, due it being ineffective and almost unusable for many desktop users. Many potential GNOME 3 users have been opting to go with KDE, XFCE, MATE, and other environments instead of GNOME 3, even on systems where GNOME 3 is the default environment. We've seen something similar happen with Firefox, where many of its users fled to Chrome and other browsers, thanks to all of the unwanted UI reworking and other unwanted changes. This has left Firefox at only about 8% of the browser market.
While Firefox and GNOME 3 may still have developers working on them, and even have some users remaining, their dwindling number of users means their vitality is dropping significantly. They are, in essence, causing themselves to be "killed off".
Fuck off.
1) Can someone make it very clear just what the relationship of OpenIndiana to IllumOS is?
2) How exactly does NexentaOS fit in? And NexentaStor? And StormOS? And SmartOS?
3) At least several of those I mentioned are open source/free, and I believe there are others. Why so many forks? Which one looks like the leader?
The product formerly (freely) available as OpenSolaris had a lot to recommend it. FreeBSD has been playing catchup and has come a long way, but is still lacking in various ways. Linux is an excellent product, but glaring probems exist in the direction it is going, and I don't see it ever coming close to matching the OpenSolaris feature set in my lifetime.
Does it come with a Taliban-style beard and thick black glasses?
I don't normally get into OS wars nowadays, but in this case, I agree w/ AC. Not wrt FreeDOS, Haiku or Amiga, which don't play in the same space, but certainly wrt FreeBSD, Linux and Windows Server. Particularly if OpenIndiana still doesn't support SPARC.
I'm just wondering - what's the point? Are there still people out there still hung up on SVR4 vs BSD that they'd prefer something like OpenIndiana over several flavors of by now proven BSD solutions - like FreeBSD, OpenBSD, et al? And I've not even touched Linux, which dwarfs even the BSDs. The only place where I see this being an option is on healthy legacy SPARC hardware that Oracle might charge 2 arms and a leg for, and where having something like OpenIndiana enables an existing platform to fork away so that they can keep running w/o the high overhead of paying Oracle maintenance costs.
In which case, if the platform ain't there on the SPARC, then what's the point? If one is looking for an FOSS server platform for x64 iron (since OpenIndiana presumably targets server installations, and not so much workstations, since SPARCstations have been dead for a while now), then one already has choices like Red Hat, Debian, iXSystems and from Oracle itself, Oracle Linux. There ain't a compelling reason to go OpenIndiana unless the original platform is a SPARC to start w/, and one wants to do a gentle migration from Solaris to something FOSS, where alternatives are either abandoned (RedHat or Debian) or largely ignored (FreeBSD). Or if someone is still caught up in the UNIX wars time warp from the 80s - the System V vs BSD crusades, and refuses to have anything to do w/ FreeBSD or NetBSD.
Illumos is the Solaris workalike kernel itself. OpenIndiana is the defacto successor distro to OpenSolaris, like OpenOffice is to StarOffice.
Distrowatch.com and Wikipedia have articles on many Illumos-based distros. Most of them have evolved into storage (eg Nexenta into NexentaStor) or server-focused offerings (OmniOS, SmartOS). This evolution appears to have been powered as a result of the needs of folks like Joyent et al.
The GP comment says absolutely nothing about sysvinit.
So why the fuck would you even bring it up?
You do know that there are more init systems than just sysvinit and systemd, right?
The important point in the GP comment is that OpenIndiana is not using systemd.
As long as it isn't using systemd, it really doesn't matter which init system it is actually using.
Sysvinit is totally irrelevant.
It's the lack of systemd that's most important.
So don't bring it up!
I suspect it's because you're using a BSD. I don't mean that as an insult to the BSDs but to the Linux-centric cruft the Gnome developers have been adding to their beloved DE, not the least among them the hard-coded dependency on SystemD(ead).
Just stick to Lumina (hope it gets a proper Linux port) or LXDE or its even more awesome-looking QT-based successor.
I have nothing but love for old Sun OS's but can someone please explain why anyone these days would choose to run Open Solaris over Linux?
Come on, if you're going to go Solaris, at least ship it with SunView or NeWS as the window system!
RHEL.. Because it is the best and most well supported choice. Is anyone really deploying Solaris in anything serious these days? Or anything other than RHEL? I don't see the point.
So on FreeBSD, the equivalent things are jails and Bhyve, the latter of which is their virtual machine. In what way are zones superior?
Seems to me the only real advantage, but a very tempting one, mind you.
Will have to check it out.
It's that they never PROPERLY introduced low-end PC-priced hardware and software options to educate a 'next generation' of solaris admins. The only people I knew who were learning Solaris by the late 90s/early '00s were people who were either at a University with a Comp Sci/EE department full of Solaris Boxen, or working at a large company that could afford to throw around money on Sun/Oracle support contracts.
Judging by what I've heard recently, the same is true today, only even people buying Oracle DB appliances are going x86/Linux over Solaris. Not due to technical aspects but simply due to costs associated with SPARC/Solaris appliances which are now difficult to find younger (cheaper) admin to administer.
And unlike OpenIndiana they don't require ZFS to run (the OI/Solaris10+ package manager uses ZFS for snapshotting/uninstalling/reverting packages, so you apparently can't even install it on UFS anymore.).
Regardless Illumos doesn't support sparc32(Sun ripped out support for it after Solaris 9), and there is probably more sparc32 hardware floating around than sparc64 in the common man's realm. Given the limited supply of sparc hardware at afforable prices housing 4 or more gigs of ram, it seems ludicrious to run illumos/OpenIndiana on it to begin with. Doing it on whatever drives came in the systems to begin with is probably even worse (Everything up to the mid '00s was SCSI based, and good luck finding affordable high-capacity SCSI HDDs that aren't close to failing. Last I looked 7200 was the minimum speed and 10/15k spindles were standard.)
but what if they have man-buns and drink PBR?
Are you kidding? I've found that Illumos is not at all friendly to white-box hardware. Examples:
Some 2-port AHCI cards mysteriously fail after a random amount of time (oops, there goes my L2ARC device until I reboot).
AHCI hot-swap on my motherboard SATA ports is a game of Russian roulette. I'll randomly get write errors on other drives when I slot a drive in. At least hot-swap worked well on my SAS HBA.
Hardware sensors that work fine with lm_sensors on Linux are not at all usable on Illumos, which expects IPMI.
KVM on Illumos? Not if you have an AMD CPU.
USB3? LOL!
A couple of weeks ago, I finally ditched my home OpenIndiana system (was running 151a9) and threw CentOS 7 on it with ZFS from the official ZoL repo. Yeah, it has systemd. So far, I've found that systemd isn't any worse than SMF; I can live with it. Things that wouldn't work in OI work nicely on C7. The ZFS pool imported without a hitch, though I had to rework my sharenfs options and do a little tinkering to get sharesmb to work with Samba. It was worth the effort. I no longer need to use crazy ACLs to get my shares to work as expected, another plus. I might try later to run a ZFS root, but for now I'm using MD-RAID mirrors for the OS itself.
Aside from the better hardware support, a big advantage of CentOS is that I don't have to worry (as much) about something breaking when I update the system. OpenIndiana's new Hipster approach is fine if you're the type who would run Debian Sid on a server, but that's not my style. Add to that, there's no clean upgrade path from 151a9 to Hipster.
I wish them well, but my foray into Solaris-land is over.
Oh, no! You have walked into the slavering fangs of a lurking grue!
XStreamO
Well, some say that the Solaris kernel is superior to Linux. Read what long time Linux kernel developer Con Kolivas says about the difference between OpenSolaris and Linux:
https://en.wikipedia.org/wiki/Criticism_of_Linux#Kernel_code_quality