Monday, January 31, 2005

Journalistic Nuggets

In Sun opens up Solaris 10 code, we get

As predicted by vnunet.com, Sun Microsystems has officially opened up the source code for its Solaris 10 operating system.

Some prediction. Sun have been shouting this from the rooftops for months, so trying to take the credit seems overly keen.

Meanwhile, HP's Unix strategy 'still needs work'. They have one, right?

Mind you, the Linux market to exceed $35bn by 2008. Boy - and it's supposed to be free?

(And let me remind you that Solaris can be free as in beer and speech.)

Going back to big numbers, I was intrigued by the snippet in an InformationWeek article:

There are about 300,000 Solaris developers worldwide.

Now, I know we've been arguing all along against this stupid idea that we don't exist, but I'm not sure even we would claim that many developers! Where did they get that number from, I wonder?

Technorati: - Technorati:

Wednesday, January 26, 2005

OpenSolaris alive

So OpenSolaris got officially announced last night.

In the pilot, we're charging ahead. As evidenced in some other places, we have source, and are building it, and running OpenSolaris on our machines.

Hooray, it works!

Sun Microsystems Inc. SunOS 5.10.1 tribble Jan. 26, 2005
SunOS Internal Development: ptribble 2005-01-26 [tribble]
bfu'ed from /peter/tribble/archives/sparc/nightly on 2005-01-26
Sun Microsystems Inc. SunOS 5.10 Generic January 2005

% uname -a
SunOS peter 5.10.1 tribble sun4u sparc SUNW,Sun-Blade-100

Technorati:

Tuesday, January 25, 2005

Sun starts to roll out OpenSolaris - Computer Business Review

A nice article: Sun starts to roll out OpenSolaris - Computer Business Review which makes some important points.

It's also balanced. Regarding Sun's CDDL vs. the GPL:

Obviously, many people like this viral nature, and many people do not.

There's room for both, and anyone should be free to choose whatever license suits them.

We're mentioned:

The exact nature of the OpenSolaris community is a bit of a mystery. Goguen [Sun VP] says Sun had already established a pilot community, which will be the core of the community to start.

We've been working on this for months. It's been fun. It's still fun. I think it's fair to say that the community being established on the pilot is pretty varied.

Simply sign up and you're in.

Unfortunately it's not quite as simple as that. As I mentioned, Sun have put together a decent bunch of people with a wide range of interests, and we've all had to sign an NDA, and Jim Grisanzio tries to keep us in some sort of order. And true believers get in on the pilot as well. Look at the blogs listed on the opensolaris.org site when it goes live and you'll see us amongst the Sun folks.

"The goal is to do Solaris development in a community-centric way," says Goguen, "but not to compromise the quality and testing that goes into commercial Solaris."

Absolutely. One of my core requirements is that the core values of Solaris not be thrown away. That's not to say that the process isn't going to involve change, or that people outside Sun won't be able to do things with OpenSolaris that they couldn't do previously.

Analysts: Paid too much?

According to eweek: Analysts: Sun's Open Solaris Plans Face Problems.

The problems mentioned are largely that of community building. And that community exists already, and is building up and getting stronger all the time. These analysts who say that Sun can't build up a community simply don't know what they're talking about. We already exist.

Webbink [Red Hat's General Counsel] agreed. "It's not about the license, it's about the community," he said. "So how is Sun going to instantly attract hundreds or thousands of developers to Solaris when they have never had the opportunity to work with the source code before?"

The following numbers are from memory, but the general magnitude is what counts. Well, Sun have about 1000 of their own developers (not all involved with the kernel, of course) already, with getting on for 100 of us external participants in the OpenSolaris pilot community. Red Hat have 200 developers? The Linux Kernel development community isn't that large either. And Red Hat have had their own problems in building a community.

So the point? OpenSolaris has already started to build a healthy community, and it's already getting to be a decent size.

Saturday, January 22, 2005

Bad Nerd Score

There has to be a bug here.

James Dickens is 92% nerdy.

Mine:


I am nerdier than 97% of all people. Are you nerdier? Click here to find out!


As I say, there just has to be a bug here somewhere!!!!!!!

No, no, I repeat no! I'm not!

Java Kstats

One of my little projects at the moment is Java access to Solaris kstats, using JNI to interface with libkstat. The idea is to bring Java up to parity with perl in this respect - and the Sun::Solaris:Kstat module h(along with the associated kstat command) has been very useful, limited in that it doesn't naturally do graphics (and there are cases where a picture is worth a thousand words - or numbers).

OK, so development has been painful - I've goofed up several times in the code. And every time I goof up java dumps core on me. (Using JNI means that all the safety built into Java goes straight out the window.) But I've moved on from simple lookups to a fully fledged kstat browser:



Technorati:

Friday, January 21, 2005

So what CPUs do I have?

There are several ways to get Solaris to tell you what cpus you have. The obvious thing would be for psrinfo to tell you, but it doesn't really:


% psrinfo -v
Status of virtual processor 0 as of: 01/21/2005 10:39:49
on-line since 01/11/2005 09:53:35.
The i386 processor operates at 2000 MHz,
and has an i387 compatible floating point processor.
Status of virtual processor 1 as of: 01/21/2005 10:39:49
on-line since 01/11/2005 09:53:39.
The i386 processor operates at 2000 MHz,
and has an i387 compatible floating point processor.


OK, so you can grope through the output from prtpicl -v, or you can use (at least on Solaris 10 where it works great) kstat cpu_info:::brand cpu_info:::implementation. I've written a cpuinfo script that tells me exactly what I want to know:


% cpuinfo
CPU 0:
AMD Opteron(tm) Processor 246
x86 (AuthenticAMD family 15 model 5 step 10 clock 2000 MHz)
CPU 1:
AMD Opteron(tm) Processor 246
x86 (AuthenticAMD family 15 model 5 step 10 clock 2000 MHz)


That's better than just a generic i386, isn't it?

Technorati:

Wednesday, January 19, 2005

Register: Solaris for hippies

According to the Register: Sun's Solaris for hippies to arrive next week.

Hm. Right. So we're "hippies".

Mind you, there are insights such as

opening up Solaris provides a way for Sun to get advocates of the OS to work for free.

Might be true, but I prefer to think of it as allowing people outside Sun as being able to help improve Solaris.

Quaint or antiquated?

It's the 21st century for Heaven's sake!

And yet some of our operating systems have some features that could be described as quaint.

For example, the regular mail client doesn't understand mime; we still have some shells that don't do command line editing; some operating systems think that compiling a kernel is a good thing; monstrosities such as rsh, rlogin, telnet, rexec still exist.

Another entertaining one is the fact that many operating systems still ship with uucp networking. Oh good grief! I've been using these things for almost 20 years and I've never come across one in all that time that's ever needed it. Are there any uucp capable systems still out there?

In fact, I think I was being too generous. This sort of thing isn't quaint. It's antiquated, and we need to move on and do better.

Monday, January 17, 2005

Development communities

I recently read either a copy of Linux now a corporate beast or something similar. (It wasn't that one exactly, but either repeated elsewhere or a different article making the same point - that the Linux [kernel] development community is small and corporate.)

The interesting thing is the following:

“People’s stereotype [of the typical Linux developer] is of a male computer geek working in his basement writing code in his spare time, purely for the love of his craft. Such people were a significant force up until about five years ago,” said Andrew Morton, whose role is maintaining the Linux kernel in its stable form.

Morton said contributions from such enthusiasts, “is waning.” Instead, most code is generated by programmers punching the corporate time clock.


Bother.

OK, so I'm male, but wouldn't quite call myself a geek. I don't have a basement, but my involvement with is definitely for the love (and fun) of it. I'm definitely not punching the corporate time clock on this one.

Of course, even with OpenSolaris being opened up, I would expect most of the changes - for quite a while - to be from Sun employees. Those same employees who are writing Solaris at the moment. I'm just hoping to be able to find some small way to contribute.

Thursday, January 13, 2005

Fedora facing challenges, just like Solaris

I was looking at the CNET article Red Hat tries again with Linux enthusiasts, regarding Red Hat's attempts to engage the wider developer community with Fedora.

On the pilot program - and one can make a reasonable analogy that Fedora has a similar relationship to RHEL as OpenSolaris does to Solaris - the challenges are similar. Sun are trying to engage with developers and others in order to build a viable, thriving community. And it's working. But it's a hard, slow process. It's a lot of work, because OpenSolaris wants to avoid the problem Red Hat had with Fedora, as Greg Dekoenigsberg, Red Hat's community relations manager, says in the above article:

"There just wasn't much they were able to do," he said. "(This time) we want to make sure we have systems and processes to make sure these people can contribute."

With OpenSolaris, Sun are trying to get this right first time around. We don't know how exactly, or how well it will work, or whether it will all be done and dusted at the time OpenSolaris comes out (clearly, there's a choice to be made between having all these issues sorted and getting the code released as soon as possible), but it's something that's important.

Sun's open-sourcing of Solaris has been criticized by some on the grounds that it will be difficult to build a community. Sure, it's going to take time before any of us get to the stage of being able to commit changes back in, but there is a community there ready to take on the challenge.

Sun have an advantage over Red Hat and Fedora, actually, and it's mentioned in the article as well:

"They (Red Hat) now are really viewed as the big commercial company," Haff said. "They can probably over time increase the user community involvement to some degree, but things like Gentoo and Debian are more natural places for the community to get involved."


Well, there currently is only one Solaris, so this dilution of effort into different distributions isn't an issue. Not yet, anyway. Who knows where OpenSolaris is going to end up?

Wednesday, January 12, 2005

Enterprise Unix Roundup: The Evolving OS Landscape

Another one from ServerWatch - Enterprise Unix Roundup: The Evolving OS Landscape

Solaris 10 is free, a move designed to pit Sun directly against Red Hat. But by making its operating system free for everyone from home users to Fortune 500 enterprises, Sun presents a compelling alternative.

Not only is 10 free (as in beer), it'll soon be free (as in open source). It's fast. It's solid. It's stable. And it's got features to drool over.

The timing on Solaris 10 and NetWare 6.5 couldn't be better.

The timing is a reference to the demise of NT4 support, but the interesting time for me is the end of January when we're expecting the real Solaris 10 to hit the streets. (As opposed to the beta/Solaris Express that we've been running for the last year or more.)

Friday, January 07, 2005

Solaris in consumer press

So I get back from the supermarket with my copy of Computer Shopper, sit back on the sofa, and open it at a random page, to be greeted by the headline:

Sun releases Solaris 10 for free

I expect this sort of thing in the trade rags, but it was interesting to see it in a consumer oriented publication.

(For reference, it's on page 333 of the Feb 2005 DVD edition.)

This article isn't on their website, but there is another related article about Sun's new Open Source License, the CDDL.

Thursday, January 06, 2005

Solaris 10 Gets Ready to Roll

So ServerWatch have noticed: Solaris 10 Gets Ready to Roll.

They get some things spot on:

To its credit, Solaris 10 has features no other operating system on the market can claim. That's right, Sun Microsystems is innovating in the operating system realm again, and it is doing it well.

and

Amazing performance and support for x86 and AMD64 platforms; Innovative features in both performance and usability categories.

Some things I'm not so sure about:

An abundance of innovative new features, mostly aimed at administrators rather than users

The abundance of new features is fine; but the Java Desktop System - Gnome, Mozilla, Evolution, StarOffice and the like - seems aimed squarely at end users. Or:

Solaris 10 is a viable option for all servers

It's close - hadware support is improving, and most (but not all) servers are well covered. Or even:

Solaris 10 comes with true 64-bit AMD support, which results in a three-fold speed improvement out of the box

Now, I know it's fast, but that factor 3 is a bit arbitrary. Some things in Solaris 10 - especially on Opteron - are spectacularly quick. Others are just OK.

And some things are just plain wrong:

Version 10 is different, however, because it uses only one code base.

Nope, sorry. It's been that way since Solaris 2.4. Or:

Cannot coexist on the same physical drive as another operating system.

Which is simply untrue. See http://multiboot.solaris-x86.org/.

Wednesday, January 05, 2005

HP Embraces Open Source as Well as Linux

James Dickens has an imaginative view on this story: HP Embraces Open Source as Well as Linux.

(Although the title would seem to indicate that HP don't see Linux as Open Source?)

On reading the article:

"we've made Linux one of our three key operating systems: Windows, Unix and Linux."

If I were Linux, I would be less than impressed at being put in that group, especially with HP's commitment to Unix. Then:

"no sizable company sitting behind J2EE."

I'm not sure Sun, IBM, or BEA would be impressed by that statement! But then:

"HP is making both JBoss and MySQL products and services into part numbers to make selling them easier to HP's customers."

Ah, right. Sell. Just a straight moneymaking ploy, then.

HP have got one thing right, though:

"tremendous need for Unix systems."

Presumably not from HP, though. They're going down with Itanic, trashing Tru-64, and going nowhere with HP-UX.

Solaris licence gets OSI approval - ZDNet UK News

So Solaris licence gets OSI approval.

Well, that's not quite right. Sun have got the license approved, but even we members of the OpenSolaris pilot program don't know for sure that this is the license that will be used for OpenSolaris. We hope so, but we'll have to wait and see.

Sun Diagrams

Just downloaded Sun's Configuration diagrams.

These are really simple - just presentations with pictures and diagrams of Sun gear that you can cut and paste into your own documents. It'll make the job of documenting what goes where in our racks a lot easier, so that job will fill in my idle moments over the next couple of days.

Oh, and I'll be using the OpenOffice 2.0 snapshot to do this.

Tuesday, January 04, 2005

Catching up

Thanks to Alan Coopersmith's Catching up blog entry for mentioning me. I just want to confirm the bit about us not being small fluffy, rapidly reproducing interstellar parasites. See some proof.

I've been trying to get one of the other sort (the small, fluffy, etc.) on eBay, but they seem to be reasonably popular and thus go for reasonable prices.

Sunday, January 02, 2005

Lifecycle thoughts

In The Care and Feeding of FOSS, Craig James discusses the Lifecycle of Software technology.

It's an interesting read, and has lots to agree and disagree with.

It also talks about Sun and OpenSolaris. For example:

There are, as of this writing, two major holdouts against the dominance of the Gnu Linux FOSS operating system: Sun and Microsoft.


Really? Sun is a major contributor of Open Source software. Including OpenOffice, an open source product that which strikes straight at Microsoft's heart. And Sun is pushing hard to get FOSS into much more widespread use, through it's (appallingly named) Java Desktop System.

Sun's case is interesting. Unlike HP and IBM, Sun has not embraced Linux. Instead, it is taking a, "If you can't beat 'em, join 'em" approach by making Solaris itself a FOSS technology. They believe the superiority of Solaris will attract talent and users away from the Linux effort. They predict that this will ultimately build momentum, until Solaris displaces Linux altogether as the primary FOSS operating system.


We might see users attracted from Linux, but don't forget that Solaris already has a strong community of its own. And some of us get very upset to be told that such a community isn't already present.

This author predicts that the opposite will happen: Solaris has many excellent technological advantages over Linux, ...


Dead right. And Linux has advantages over Solaris. Each has different focus areas. I use Solaris because it does a better job - for me - of solving my problems.

...but rather than attracting developers to the unfamiliar Solaris OS, those developers will incorporate the Solaris technology into Linux. The end result will be a hybrid, a marriage of the best of Linux and Solaris, but it will be called Linux, not Solaris. Linux simply has too much momentum to be derailed by the upstart (to the FOSS community) Solaris.


I don't think so. Some of the technology can be easily ported - in the same way that Solaris contains an awful lot of free software already, but other key technologies are based on architectural principles and are going to be a lot harder to transfer to any alternative OS. Dtrace, for example, is based on the fundamental belief that the OS should allow thorough debugging and diagnostic capability, and is intrinsically reliant on a kernel architecture based on that philosophy. Linux could, if wished, dramatically enhance its capabilities in these areas, which are significantly lacking, but it doesn't seem to be a high priority.

Maybe somebody will come up with a Gnu/Solaris distribution. It shouldn' t be that hard, especially with OpenSolaris.

I don't actually see the release of Solaris as Open Source as having that much of an impact on the Linux kernel. After all, Linus has said that he's not going to bother taking a look. And the codebases are likely to be sufficiently different that you can't simply copy the code. Taking the fundamental design principles of Solaris and using them to enhance Linux would benefit Linux, but you don't really need the source to do that - you just need to understand what those fundamental principles are.