27 Jul 2000

Thu, 27 Jul 2000 20:29:43 GMT

Hi, DeWitt. It's been awhile.

If anyone knows of a nice cheap place to live in Cambridge (or is looking for a roommate), let me know. Moving is a hassle.

9 May 2000

Tue, 9 May 2000 04:37:06 GMT

I haven't yet figured out how users are supposed to be certified initially. (Is it the case that everyone knows someone else already here?) Well, maybe someone will read this, take pity on me, and call me "apprentice." I'll even post qualifications (thanks to the odd new Orbiten Free Software Survey). This is me: http://orbiten.org/ofss/codd-render.cgi?action=author&key=1353

Not spectacularly impressive, is it? (Aside: I wish they hadn't posted my email address like that. I figure I'm a few days away from a deluge of spam.) At any rate, I have written some free software.

Now that I've presented my meager credentials... About Pike's presentation on systems research:

I think a lot of people here have missed his point, particularly with respect to free software. To the extent that he is, as Advogato writes, "pessimistic...about free software," his pessimism stems from its irrelevance to systems research. That, after all, is the topic of the presentation. (Successful) free software projects are rarely ever original (in the sense employed by Pike; that is to say, broad architectural strokes aimed at rethinking full system interaction. What Pike says about Linux, in particular, is not that it is a poor system, but that it is an unintertesing one; it copies an old design.

Consider his point of view, as a computer scientist (as opposed to, I mean, a systems administrator or a software developer): UNIX is old hat; how could anyone deny this? If your profession is about doing original operating systems research, how could you fail to be depressed by the fact that such research is at an ebb?

Of course, one might deny that it is at an ebb. Graydon, for instance, seems to find this notion absurd. However, one should notice that out of the many links he provides, the majority are not systems software, but programming languages. And what about the systems that are listed? How many of these are in production, or look like they ever will be? Okay -- now how many of those look truly interesting from a research point of view? The OSKit, for instance, is a tool for OS researchers and certainly not an "interesting OS" in its own right. I think that Pike has a different and more demanding notion of what comprises original, interesting systems research than does Graydon. Note that Graydon writes:

the examples I cited are innovative because they embody research and experimentation in a "systems" area which the "Unix, sockets and C" holy trinity has had some difficulty with, be that fine grained nested security, application control over hardware, transparent clustering, process migration, provability, redundancy, self-repair and self-organization, calable distributed computation, realtime, or higher order abstractions.

But this isn't systems design, as desribed by Pike. Take page 16, where he lists the (discouraging) trends in current research. Number 2 is: "Don't go for breadth, go for depth. (Microspecialization, not systems work.)" The point is that focusing on "fine grained security" (or anything else) without paying attention to how it works as part of a system with a unified architecture does not qualify as real systems work. If ten specialists focus on ten different, narrow jobs, why should anyone suspect that their results will somehow magically cohere? Well, they won't, and Pike knows this.

Advogato blog for jaz

27 Jul 2000

9 May 2000