On Sun, Aug 15, 2010 at 8:09 PM, Ray Arachelian <r...@arachelian.com> wrote: > On 08/15/2010 07:13 PM, Jonathan Edwards wrote: >> but how do we liberate the source from the CDDL? > > Why would you need to? The kernel can remain CDDL. Anything that can > be replaced with sources from *BSD or Linux directly, can be, if needed, > but why would a non-Oracle distribution of a fork of OpenSolaris need to > be non-CDDL? > >> that still seems like it could be a bit of a problem >> if the original license steward (Sun => Oracle) decides >> to change the terms of the license, and/or if the >> larger community does not trust or approve of the >> actions taken by the license steward .. >> > > If Oracle changes the terms of the license, there still would exist man > pages and include files in Solaris proper. There's nothing to stop > anyone from implementing *similar* interfaces. Sure, the binaries would > then no longer run, but the source code would compile just fine. The > back end implementation would no longer work exactly in the same way, > and if you were to run things under a profiler, some code would run > slower, other code would run faster than on Solaris proper. But it > would run. So in a sense, it would be source compatible, at least. > > Anything patented would become problematic, so one would need to watch > what patents are registered and work around them, or do the exact > opposite and avoid looking, whatever the best strategy is. > > But if Oracle does go belligerent in that manner, there's not much to > stop it anyway, CDDL or not. Does the existing CDDL'ed open source code > in osol provide a "will not sure for patents covered by this source > code" clause? > > Key is, Oracle cannot retroactively change the license for already > released source code. If there are no possible patent issues to the > already released CDDL code, then all will be fine, except that binaries > from Solaris 11 and future releases might not run properly on future > IllumOS or Nexenta if the ABI changes. > > Would it be so bad if IllumOS was forked off Solaris? IMHO it would be > more similar to FreeBSD, OpenBSD, DragonFlyBSD, etc. forking off > NetBSD. Oracle did claim they'd release sources for Solaris, maybe they > will, maybe they won't, but if they do, there'll be something like a six > to twelve month gap between release and incorporation into IllumOS (to > account for code reviews, and non-Oracle code changes, as well as bug > fixes and such.) > > As IllumOS grows and gains more participants, bug fixes (I'm thinking > security fixes mostly) can be provided independently of Oracle. > > A lot of the Solaris distro itself does come from outside. Things like > OpenSSH, OpenSSL, sendmail, apache, gcc, etc. These are already open > source and widely available, so maintaining these will be easy. The > hard parts would be proprietary libraries, kernel, and kernel modules > and device drivers. > > As long as Oracle doesn't become evil (for highly large values of evil - > i.e. suing for patents), things will work out just fine. Ideas for > features as well as drivers can be mixed in and incorporated from the > BSDs and Linux. > > Perhaps some sort of side project can be started to look at a generic > way of making device driver sources from Linux and the BSDs work with > the OpenSolaris kernel (as separate Solaris modules so there are no > clashes between the CDDL and GPL/BSD licenses.) Most of the devices are > block or character mode, so perhaps some sort of compatibility framework > could be built? Linux certainly has a lot more supported drivers for > hardware than opensolaris, if there was a way to port a majority of > those drivers, we wouldn't have to rely on Oracle for them, nor would we > have to rewrite them all from scratch if there was some compatibility > layer. Granted not everything would work immediately out of the box, > but for stuff like USB and PCI devices, as long as it doesn't depend on > too many things inside the Linux or BSD kernel, it might work. (And > those callbacks could be implemented in another layer.) I recall about > 5-10 years ago the idea of a captive driver on Linux where you could use > windows NT drivers under Linux, for example. > > If Oracle doesn't open source certain aspects of Solaris, perhaps the > IllumOS project can provide similar projects written from scratch with > identical commands and similar or identical arguments for them. For > example, flash archives aren't provided, but new independently written > installer that supported them, along with a clean room written flar > command set could be written as well, etc. > Or hell, grab kickstart and tweak it and work from there until a > comparable installer that does as well as or better than jumpstart is > produced. > > And it could also have a flar restore command that could be executed > from a live CD against a hard drive. It doesn't have to follow it > exactly as long as the wanted features and support are there. > Published information such as this: > http://docs.sun.com/app/docs/doc/816-5174/flash-archive-4?l=en&a=view > could be used to make a new implementation, even if incompatible. > > I'd suggest, whatever Oracle releases, we use to keep compatible with, > whatever they don't, we implement from scratch with comparable > features. After a few years, distros built off IllumOS would likely > outpace Solaris, and Oracle would find themselves having to keep up with > IllumOS instead of the other way around. The difference was that most > of the OpenSolaris developers were at Sun. Now that Oracle's done a bit > of laying off, these folks are now freed of Oracle and don't have the > same limitations. If there's a far larger developer mass (assuming > there is interest) outside of Oracle than inside, this can happen. > IllumOS doesn't have to compete against Linux and the BSDs, rather it > can join them and both benefit and contribute to them as well. > > If you remember the bad old days of Solaris 8 and 9 where right after > you installed an OS, you had to spend tons of time raping the bandwith > of sunfreeware.com for all the GNU and non-GNU free packages that were > missing from the Solaris, just remember how much came to the Solaris > distro from the outside. Eventually, Sun got the hint and put a lot of > those packages under /usr/sfw. This can be a similar situation a few > years out from now if we get enough interest and momentum to outdo what > Solaris 12 for example will have, and you'll see Solaris 13 grabbing > things back from IllumOS. :) > > Let's stop grieving for Sun and OpenSolaris and redirect our energies > into bettering IllumOS. We can build something even better than what > OpenSolaris could have been. This isn't the end, it's a new beginning > with a lot more freedom. > > If anything I'd strongly suggest that whatever new features are to be > added, as long as they can be, should be outside of the CDDL, and rather > under GPL licenses than BSD or other licenses. This would prevent > Oracle from taking back source code, tweaking it and closing it. That > way, anything they'd adopt from IllumOS, they'd have to release sources > for tweaks or bugfixes. :)
My understanding is anything released under the CDDL by someone that isn't Oracle (and not contributed under any SCA), would mean Oracle would have to release any changes to those files as well. They only get the right to withhold stuff they own the copyright to. What would be more interesting would be if all the cool stuff was happening in Illumos, which forced Oracle to follow it (instead of the other way around) :) > > For example, I'd love to see k-splice like features in Solaris, but that > will likely never happen, but it might under IllumOS. Want to build a > router or firewall with IllumOS, now you can. Want to build a MythLumOS > distro, now you can. Want to build IllumOS for a tablet, or hell, for > that matter your cellphone, instead of going with Android, now you can. > Want to bring back the PPC port, or even better, want to port to > Alpha-AXP even though that's now a dead platform? Now you can. Just as a side note, Solaris _8_ had k-splice like functionality (so really Linux is years behind in this sense), however my understanding was it was a support nightmare (as now the bits in memory and on disk might not match, plus the number of possible running configurations exploded into an extremely large number), etc. All of which tended to negate any benefits one might get (especially now with beadm + fast boot support), so it was abandoned. _______________________________________________ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
