Dæmon's Advocate

Poul-Henning Kamp <phk@phk.freebsd.dk>

Old junk

I remember when the Alpha chip was first unveilled in sufficient detail that we could try to understand what kind of beast it was. And I remember seeing and installing the first one, and doing some heavy duty math on it to see if the performance was interesting.

Then we heard the NetBSD crew were porting to the Alpha and we got a bit envious about that. But since none of us had Alphas we could play with much less time to do so, it did not take long before we realized that we could lift 64-bit cleanup patches from NetBSD rather than have to do the hard work ourselves.

And then Doug Rabson gave us our own Alpha port and all sorts of "trivial" things became problems. We needed ports for Alpha, we broke Alpha builds because we did not test on the machine in the cluster. Gradually, Alpha taught us most of what we know today about being a multiarchitecture project.

And now, due to random acts of management in DEC^WCompaq^WHP rather than actual architectural weakness, it has been taken out of the harness and put out to pasture as a "Tier 2" platform.

Being Tier 2 does not mean that it will not work however, only that it will not be show stopper for our release procedure. In all likelyhood Alpha will keep working as long as we can find machines which have the capacity to run -current. My Alpha machine will stay where it is, if for no other reason then because of the combined back-breaking potential of its bulk, weight and the stairs into my basement.

In other news, it looks like ibcs2 and svr4 emulation are still sufficiently desirable that they will not be stuck in the attic yet.

Anyway, that was only an excuse to get to the real topic of this column which is old junk.

Let me make a straw-poll: how many of you out there still have a Commodore, TRS-80 or Zinclair stuck in a box somewhere?

Right, me too.

It's not for myself though, it is only so I can show it to the kids when they are old enough to properly appreciate it. Same thing for my punched cards, mag-tapes, Sinclair Mk 14, and the homebrew Z-80 system. Not to mention the 386SX which will take less than 15 seconds from power on until there is a login prompt from "386BSD 0.1-newer".

Is it just me, or do we hackers have a really hard time throwing out old stuff? From what I hear people tell me I am not the only one with a problem in that department. Heck, some people even flaunt their stuff on the net: The Mike Smith memorial room. I could appear to be a deeply ingrained trait of hackers to be uncompromising collectors

But the sad truth is that despite the numerous private collections, at least half of the history of computing is rapidly disappearing: some museums have started collecting computer hardware but almost universally the software is neglegted or even ignored. What good is it to preserve and display the computers as hunks of cold old metal if we do not also get the chance to appreciate their soul in the shape of the software which ran on them?

It is hard to get this point across to museum curators. A plough is a good exhibit even without the horses and the farmer, but a computer museum without the software is like a library which has kept only the covers of the books.

I have taken the consequence of my addiction to old computers and joined Danish Society for Computer History. Doing so I instantly doubled the size of the number of members younger than 40 years of age and prompted the creation of a software interest group.

The other half of the youth fraction have already completed a simulator for the Gier computer which allows people to gain first hand experience with the computer on which the original ALGOL-60 compiler were written.

One reason to care about this is that this is the computer on which Brinch-Hansen did his seminal work on operating systems message-passing and concurrency. The Amiga crew at Commodore had read their classics, and since DragonFlyBSD is persuing message-passing I would be surprised if they have not also spent time reading how Brinch-Hansen proved his OS was free from deadlocks.

Two other reasons for caring are so obvious that we usually miss them: software is fascinating in itself and that the last 50 years have seen the world shaped more by software than by anything else.

So please make sure that your old junk is not missing in a collection somewhere before your wife forces you to clean out a room for the kids.

The BSD community is lucky because a lot of our history has been preserved in the shape of Kirks CSRG CD set, and the broader UNIX software history is also in good hands with the The Unix Heritage Society not to mention our own CVS repository. We even have historically advantaged people like Peter Salus have already started to write books about us.

But if we go back to the present, we must also realize that we cannot keep all software alive forever, and thus it will eventually come to pass for the Alpha, ibcs2 and svr4 support in FreeBSD which will at some point in the future be dropped entirely and relegated to the CVS Attic.

But it will not be forgotten.

Poul-Henning