The Nexenta Core Platform marches on with the release of version 3.0 of the OpenSolaris based distribution, based on b134. It has also officially unveiled plans of moving from OpenSolaris to Illumos, the fully open branch of OpenSolaris. It will also now plan on moving to a new userland release. Grab the iso here.
The first time I used nexenta I was a little surprised that it was such a bad dekstop OS since it was based on Ubuntu, and Ubuntu is a very smooth desktop OS. Of course, the desktop is not Nexenta’s target audience, but I hope this will change in the future (maybe by collaboration with StormOS).
Another problem is that the distribution would need to integrate many of the Solaris driver management utilities that are found in OpenSolaris, but not in Ubuntu. One big example is the wifi stack.
Yeah… Really, Nexenta as a desktop OS is …well… how do you say`? Clumsy?
Naw, we will see what happen.
Isn’t Nexenta the main sponsor of Illumos? So, maybe NC will never be targeted for desktop user, or?
Besides: I do not knwo, how StormOS can help, since they are changing from Ubuntu to Debian Userland.
Nexenta may also change to debian userland in 4.0+ I believe (check the URL in the story).
I’ve read their announcement and I am confused, do they use the Solaris ON distribution – that is, the libc implementation for example or does it pretty much grab the kernel and throw the GNU user space on top? I’m getting confused messages as to how they’re going about it.
Nexenta uses OpenSolaris kernel and libc they then use GNU/debian userland (GNU coreutils and apt-get) it’s kinda equivalent to way the now deceased OpenSolaris/Indiana distro included a /gnu/bin path containing GNU coreutils.
The first thing I always changed in Indiana was the default path which pointed at /gnu/bin (GNU coretutils do not support stuff like ZFS ACL’s etc.)
The fact that they do use the OpenSolaris LibC cause a ruckus in the Debian community, people basically said there were breaking the GPL by distributing a system with GPL software linked off a libc that was licensed under the CDDL. (FSF thought otherwise)
How is it breaking the GPL by linking to libc? I’d love to gear what evidence they have besides just sour grapes!
Nexenta only targetted desktop in it’s early beginnings. It didn’t do any Desktop since then.
There is a distro build on Nexenta that does target the desktop called StormOS.
I really would love to love Nexenta in similar fashion to Debian/kFreeBSD.
But sometimes it does still feel rough around the edges.
tbh I wouldn’t say Ubuntu was a particularly smooth desktop OS either. But then that’s my personal bias speaking more than impartial fact.
As for Nexenta, I ran CP1 for about 6 months and I really tried to like it but ultimately found it hard to love.
I’m sure the OS has come alone leaps and bounds since v1, but now that FreeBSD supports ZFS and VirtualBox, there is no need for me to run a Solaris derivative any more.
I think when they say moving to a new Userland. I think they will be moving to solaris proper userland with apt like package management system. Considering that they are working on Illumos i strongly believe that this is what is going to happen.
Prudhvi Krishna Surapaneni
I’ve been tempted by Solaris in the past but brutally disappointed each time I’ve tried it for a variety of reasons – poor hardware support being one of them along with poor performance on this like FAT file system support and so on. Funny enough I’ve tripped over this website:
http://www.freelancer.com/
And wondered maybe in the future I should advertise a whole list of pet peeves I have with Solaris and pay freelance developers to fix them up – I know I can’t programme for donkey balls but I wonder if I put up the money to pay for some of the problems to be fixed – what would the ‘Solaris community’ think of it? I’ve mentioned bounties in the past on the mailing list and chat room with the result being the cold shoulder by some regarding it.
I’d love to hear some feedback by developers regarding this idea; some of the ideas I have at the top of the list includes paying for an OpenBSD wireless layer so that wireless drivers can be easily compiled from OpenBSD to Solaris, maybe fund a project that totally removes HAL and replaces it with something else, improving the power management, support the full/complete implementation of the usbvc standard so that more web cams can be fully supported and include not only the latest CUPS but latest drivers from gutenprint and other driver repositories (and do not require more tweaking after installation as so many times on Solaris setting up printing requires) so that they can either be download and installed easily as binaries or included in the community distribution.
Edited 2010-08-21 10:14 UTC
OpenSolaris/Illumos uses the FreeBSD wireless network stack. This is shared with both NetBSD (where it had originated) and OpenBSD (since 3.6). Several wireless drivers have been ported to OpenSolaris from both FreeBSD and OpenBSD (rum for example)
This wireless stack (net80211) has also been ported to Haiku as can be seen in the article here:
http://www.osnews.com/story/21822/Haiku_WiFi_Stack_Prototype_Connec…
There is also a project to port it to Mac OSX. Having a common Unix architecture for generic ieee80211 layer is a good idea in my opinion.
net80211 originated in NetBSD from what I recall before been adapted by FreeBSD and subsequently improved. There was an attempt to bring it into Linux but it was seen by some kernel developers as been “tainted” due to it’s *BSD origin.
Generally I find the best support on laptops comes from those that are labeled as “Centrino”
Cool, the problem is that many laptops these days are being bundled with Broadcom wireless especially those of the NZ$1000 range; there is the bwn driver from OpenBSD and the Linux STA driver from broadcom, neither one has been ported to Solaris even though the hardware is hugely popular out there. Yes I understand I can use NDIS but it isn’t a particularly good solution given how iffy it is in terms of stability and the idea of running Windows code on Solaris does send a shudder up the spine. Hence my wondering whether I’m better off putting up a one off contractof $1000 to pay for Linux STA driver or the bwn driver to be ported to Solaris.
I know this is unrelated to your post but I’ll tack it onto the reply – As for power management, the last reply on the mailing list for the tesla-dev was in June, has nothing happened in terms of development since then? kind of depressing that Solaris is sucking down the power and nothing is being done about it. I remember that there was a move to make the kernel tickless with the deprecation of lbolt64/lbolt in favour of ddi_get_lbolt()/ddi_get_lbolt64() and yet new drivers are still being merged that use the old deprecated calls – its a mickey mouse setup if you ask me! As for the tickless project, it is in a state of riga mortus since March with no new mail on their mailing list since then. To be pretty damn honest it is depressing to say the least.
But then again I can hardly complain if I fail to put my money where my mouth is and actually do something about by either paying someone to do it or address it myself. I guess in terms of such development there are higher priorities such as finishing off i18n in libc and other components that need open source replacements.
Edited 2010-08-22 02:17 UTC
Well I think the illumos guys at the moment are working on just having a booting system. It just shows though that Sun could have easily replaced the closed bits of libc by tasking 1-2 engineers to it. Garrett basically replaced it in a couple weeks of work all by himself. Perhaps Sun preferred that certain elements were kept closed source as it force the maintenance of an umbilical cord where 3rd party distro’s were dependent on Sun to continue releasing the closed binary bits with each update.
Regarding Broadcom, I would think part of the issue here is no one has raised a bugfix/change request to incorporate bwn. If it was raised and relevant links pointed to OpenBSD/FreeBSD you might get a quicker turn around. (Emphasis on *Might*) Illumos do have a feature request list, I see for example they’ve added extra support for different Marvell Nic’s to yge.
I myself refuse to buy laptops with Broadcom chips in them since I had a similar issue with Linux back in 2004. My solution was to buy an Intel mini-pcie wireless card (2915abg) card and install it. Result I had a Athlon64 laptop with intel wireless 🙂 which worked out of box with whatever *nix I put on it.
True, I always got that impression – give the community enough freedom to think they’ve got some autonomy but not too much or otherwise they’ll run off with the code and create a competing product. Add on top of that the laborious procedure just to get anything accepted as a new feature/enhancement/bug fix, I couldn’t help but come to the conclusion that OpenSolaris nothing more than a giant marketing operation rather than a genuine open source project.
I’ve made bug reports in the past and they’ve been ignored in the past – there is nothing much one can do if OpenSolaris developers ignore bug reports.
Today it isn’t so easy given that many laptops apart from maybe Toshiba, Lenovo and the rare HP laptop, are using non-Broadcom wireless chips. One can avoid trying to get laptop with Broadcom or just bite the bullet and support it. At the end of the day one shouldn’t need to jump through hoops to get support nor does everyone purchase a laptop with Solaris in mind – some purchase it with Windows first, they see Solaris and want to migrate across – why not make life easier but supporting as many device as practically possible?
Edited 2010-08-22 12:01 UTC
Is a lack of a detailed procedure to use latest pkgsrc. I think Nexenta could provide this.