Posted on 09/15/2005 7:29:08 PM PDT by Halfmanhalfamazing
Despite the obvious driver problems experienced on commodity x86 hardware, there is not much else to criticize about Solaris 10/OpenSolaris. There are some obvious gaps which should be addressed over the long term--better updates and patching processes, for example, and an improved administration framework would go a long way to help new users unfamiliar with the Solaris environment. Is it a viable alternative to Linux? Absolutely. In the last six months I've had no problems with Solaris 10 crashing, locking up or exhibiting odd behaviour. By comparison, my Gentoo-based systems have not been so well behaved. A SPARCserver 20 running Gentoo simply freezes after about six days and requires a hardware reset. My PC-based Gentoo installation often freezes if there's an NFS issue on the network and I have to reboot it.
I use SuSE 9.3 Linux and WinXP Pro on my home desktop, and both have their problems. Both are as secure as my knowledge allows, and I don't spend as much time getting to know either one as I should. The lions share of my experience with variants of UNIX is on the server level, which is where my earlier statement comes from. On the server level, it's hard to beat Solaris on Sparc. My system of choice when deploying database however is AIX on pSeries (formerly known as RS/6000). Very solid.
I stopped supporting SCO Unixware on my worksites in 2001. I was never so happy in my life as when they were removed as a customer offering.
That's a shame.
I consider myself an AMD person for price/performance reasons, but as soon as I heard that intel had some sort of plans to bring itanic to the desktop I've been one of those who wanted it. The FPU performance is unreal and this type of architecture would really have promise.
I've always thought that it's time to wipe the slate clean but every day it gets harder and harder.
First Sun, now Apple.
*sigh* x86 will never die.
export X=Y
works on every variant of Unix except Solaris.
Yeah, both systems do have their problems no question there.
You'll always give up something when going from one system to another.
> I've always thought that it's time to wipe the slate
> clean but every day it gets harder and harder.
No harder than it's been for the last decade. It "merely"
requires a CPU with over twice the performance of the
next fastest thing, and a credible roadmap for sustaining
that lead for at least five years. Itanic never had that.
Something could come along that does.
> First Sun, now Apple.
Sun may survive being assimilated.
I do wonder about Apple, tho.
> *sigh* x86 will never die.
Nope, but today's AMD64/EM64T would look pretty clean
(and somewhat alien) to an i8008 programmer. Plus,
with most new code being developed in high level langs,
any remaining awkwardness in the architecture is no
longer a real issue.
The limits that x86 placed on end users are also gone,
such as maximum RAM and IRQ count (still only 15 IRQs,
but IRQ sharing on PCI/PCIe and multi-device busses
like USB and '1384 make it irrelevent).
About the only remaining question for x86 is BIOS vs EFI.
> Something could come along that does.
No sooner penned than this news popped up ...
Ex-Alpha designers set to unveil first CPU
http://www.theinquirer.net/?article=26214
The PAsemi site itself is rather content-free, so I guess
we'll just have wait and see what comes out of their oven.
That they call themselves "P.A. Semi" is odd, since "PA"
is usually understood to refer to hp's PA-RISC CPU, the
PA standing for Precision Architecture.
Well, the kernel is simply another program. The performance hit if it's badly written is heavier than if it were userland, and segfaults become deadly inside there, but it's not "magic". It's like rewriting anything else, only with extra responsibility.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.