Solaris “Nevada” build 70 also fails to install: I give up

I tried installing Solaris Express Community Edition, code name “Nevada”, build 70. And this time I got a different error, which ended with the following statement:

Error 28: Selected item cannot fit into memory

After a bit of searching, I have discovered that it is likely caused by an incorrect build of GRUB. If I am correct, this will require rebuilding GRUB and creating a new ISO altogether. Translation: I cannot install this build of Solaris Express, at all, no matter how much I stamp my feet. I have to wait for build 71 to be released. But I am not going to.

It is becoming clear to me that older systems, such as mine, are to OpenSolaris developers what bugs are to windshields. They clearly don’t care, or it is so low on their priority list that they don’t even bother testing builds on these kind of systems. It may be worth noting that, yet again, it worked like a charm on my Athlon XP system.

This is ridiculous, are they just making it up as they go along? I know they make no guarantees regarding stability and such, but I am basing my choice on the HCL and the system requirements. That means I do not expect it to fly on my slow laptop, but it should at least install.

Why waste my time with incorrect minimum system requirements and outdated HCL entries? Or are these purely theoretical system requirements, for arm-chair computer enthusiasts to run on imaginary systems?

If this is the future of what SUN is touting as the “world’s most advanced operating system”, then I do not predict a very bright future. At least, it will be quite some time before I even think about trying OpenSolaris again.

OpenSolaris: no Pentium II support?

As I wrote earlier, I have been thinking about trying OpenSolaris. I checked the hardware compatibility list on SUN’s website, and my Toshiba Tecra 8000 was reported to work fine. The system requirements indicated that my laptop, with 256 MB RAM, 400Mhz Pentium II CPU and 40GB hard disk drive, should be able to run OpenSolaris just fine.

So I downloaded OpenSolaris “Nevada” Build 69 and burned a copy of the first CD image. Tried it on my Athlon XP machine, and it booted just fine. So I took it home and tried it on the laptop, but then I got the following error that repeated itself over and over again:

WARNING: init(1M) exited on fatal signal 9: restarting automatically

I googled the problem, and discovered that it is related to my Pentium II and its lack of SSE support. I also found the following bugreport: “snv boot failure since snv_66, no support for systems without SSE?”

So people with pre-pentium III CPUs are out of luck with builds 66-69, inclusive. Fortunately, it seems that a fix is to be implemented in “Nevada” Build 70, which I noticed became available for download yesterday.

I will try the first CD of that build later this evening and see if that works.

Thinking about trying Solaris …

When I was studying at the university, we used Solaris (up to version 8) on old Sun SPARC workstations. Oddly, at the time I thought CDE was quite a pleasant environment to work in. Though, due to the limited processing power -these were rather old systems- I was usually *forced* to use with WindowMaker instead. Since then, Sun have made Solaris available to x86 architectures and has started the OpenSolaris project.

So finally, I can try running Solaris on my own computer at normal speed. But I do have some doubts …

For example, I rely on a wireless interface to connect to the internet. While the OpenSolaris project are now developing some kind of auto-magic network managing daemon, it is still quite experimental and accordingly marked “Volatile”, which means it’s configuration can change drastically from the current version to the next.

And then there is the lack of software. The whole point of running an operating system at all, is that it enables you to use your hardware and software to be productive and watch porn. It is merely a means to an end. But when I search for Solaris x86 binaries for MPlayer, PAN or WINE, I find very little, or only SPARC binaries, or nothing at all. At least, nothing on the official websites.

There are, however, some third-party packages you can download from personal homepages. But frankly speaking, simply downloading unsigned packages from third-party websites is not something I prefer, for security reasons. That means I’ll probably need to compile these programs from source, which means reading all the README and INSTALL documentation provided with the source. It means a lot of work, and I don’t want a lot of work.

So does that mean I won’t try it? I’m still not sure. I admit I am very curious, but the lack of a desktop userbase as large as GNU/Linux’s also means you will not have nearly as much luck finding answers via search-engines. And *that* is a very, very big problem. For example: just enter “solaris ipod” in google, and then “linux ipod”, and compare the number of results. Solaris gets around 2.4 million results, whereas Linux gets over 97 million! (By the way, there is no out-of-the-box support for the ipod on Solaris. It apparently just mounts as an external disk, but software like rythmbox has no access to its database).

I’ll have to think about this…