I recently ran across Xhomer, a simulator for the DEC Pro 350. While there are several historic machine simulators out there, Xhomer is polished: it compiled and ran on my Opteron laptop running Solaris 10 with no difficulties. But best of all, Xhomer includes system software: you can download a disk image of VENIX 2.0 -- a "real-time" UNIX variant from VenturCom. Here's a screenshot of me logging in.

While I had heard of VENIX (a System III derivative), I had never used it before today. (And nor, presumably, have the good folks at pharmanex.) In using it, it's clear that VENIX had some BSD influence: VENIX 2.0 includes both csh (blech) and vi (phew). Using a twenty year old UNIX is a strange experience: I'm amazed at how little the most basic things have changed. There was very little that I didn't recognize ("em1" anyone?), and I was familiar with all of the tools necessary to write a program (vi), compile it (cc) and debug it (adb). Using the latter of these was the most amusing; here's a screenshot of me using adb.

Compare this output with the output of "$a" on adb, and you'll see why this got me excited. (And then try "$a" on mdb for our warped idea of an easter egg.) It should go without saying that I looked (so far in vain) for an Algol 68 compiler on this system. Seeing adb spit out a true Algol stack backtrace would be like sipping from the debugging fountain of youth...

While it's amazing how familiar VENIX feels, I'm also stunned by how anemic its facilites are: it has no TCP/IP stack, no real filesystem, no multiple processor support, no resource management, no dynamic linking, no real virtual memory system, no observability and poor debuggability. It reminds me how far we have come -- even before we embarked on the far more radical technologies found in Solaris 10...

Now, does anyone know of a Language H compiler for the PDP-11?

OBEY

Finally, we have the South Park of operating systems: Inside Jack. It's really pretty funny -- and it even offers a rare glimpse (albeit animated) at Mike, the shadowy and elusive third member of Team DTrace. And if you enjoyed that episode of Inside Jack, there's an earlier episode that's also pretty amusing.

So it's been an exciting week for Solaris: at long last, we officially launched Solaris 10 on Monday. Unlike most product launches, the Solaris 10 launch was heavy on both technical details and customer testimonials: it was very important to us that those covering the event understand that this isn't ballyhooed nothingness -- this is real technology that is having a tangible impact on those using it. To that end, Mike, Andy and I described the Solaris 10 technology areas in some depth to a group of fifty journalists in a Solaris "boot camp" on the morning of the launch. I was pleased by how many journalists were there to begin with, and impressed that none left over the two hours or so of informal presentations: this showed a real willingness on the part of the press to understand what we had done. (Impressively, they even stayed after I suggested to one journalist that he and I strip to the waist and wrestle to settle a difference of opinion. Fortunately, we were able to settle the difference without resorting to fisticuffs.)

But my favorite part of the launch -- hands down -- was when Don Fike from FedEx stood on the stage and described the application performance problems that FedEx has found using DTrace. It's always gratifying to see a customer achieve a win with DTrace (which of course is what motivated us to write DTrace in the first place), but it's something else entirely to have a customer be willing to stand on a stage with you and put their reputation on the line by vouching for your technology. And on top of it all, to have that customer be FedEx -- a company that I (and most, I suspect) hold in very high regard -- well, it nearly brought a tear to my eye; moments like that just don't come often in one's career...

Overall, the launch was a great success. Driving back up to the City with Mike, we wondered aloud: how would the competition respond? As it turns out, we didn't have to wait long: Martin Fink, HP's VP of Linux, dashed off a hasty diatribe against Solaris 10. As others have pointed out, this is pure HP FUD: it doesn't attack our technology in any concrete fashion, but rather attempts to put baseless fear in the minds of those who might be considering it. In particular, Fink returns to a classic FUD attack from the early 1990s: fear of a mixed-endianness planet. This was certainly a surprising angle of attack: given that this issue has been technically solved for nearly a decade, I naturally assumed that this was a dead issue for any technologist. But then, his attack reveals what is confirmed by Fink's bio (and photo?): Fink isn't a technologist. But most amusing was Ben Rockwood's hilarious response Thank you, Ben, for responding with the pluck and thoroughness that I believe characterize the Solaris community...