Silicon Valley Redneck

Kindly refrain from profiling

One baseball player crashing an airplane into a building does not constitute proof that all baseball players are terrorists.

We must resist all demands for heightened screening for athletically-enabled passengers.

Please do not shoot random strangers who happen to be wearing baseball caps.  They might not even be baseball players!  Rappers and truckers also wear such caps.

Baseball is a peaceful game, and we are not at war with it.  Only a small minority of baseball extremists are our enemies.

Paging Buck Rogers...

Continuing on this morning's keep-yer-body-fresh theme, I find that Slashdot has linked to a BBC News article claiming that sewer gas can induce suspended animation.

Well, heck, it got this guy all the way from 1920ish to the 25th century.

Ah, Wikipedia strikes again...

...this time with (reasonable) accuracy.

In the wake of the NorK Nuke test, someone linked to the Wikipedia entry on on krytrons (which turn out to be pretty much what I'd always thought they were, with a minor twist), which eventually led me to this article, which does an alarmingly good job of connecting the dots and filling in many of those little details which have traditionally been obscure (or even intentionally obfuscated).

Mind you, most people don't have access to suitable fissile materials, nor to the required fabrication facilities for making a workable nuke, so the obscurity of these details isn't really all that important; I have to assume that anyone with the resources to build a nuke also has the resources to send innocuous-seeming guys to buy beers for selected geezers and to mine their tales for useful information.  Still, I'm not crazy about having so much of it in one easily-found place.

Cryonics forecast

"Cold."

OK, so apart from that... there as a cryonics panel at Silicon, which got me to thinking.  Featured was a National geographic program which included a dissenting voice (not persuasive) and a fairly detailed account of one patient (much more persuasive, on the side of the dissenting voice).

Let's see, now: it does make sense that chilling the body quickly is a Good Idea.  People who die cold - e.g., drowning beneath the ice of frozen-over lakes - can remain dead for a significant amount of time and still be successfully revived.  However!  This typically involves an initially-healthy subject who is well chilled at the time of death.

The featured patient died of, among other things, multiple brain tumors.  This is a bad starting point.  Initial preparation was begun within minutes after she died, but that's minutes after, not minutes before (for obvious legal reasons).  I gather the body was cooled to frozen-lake temperature within an hour or so, but... this still seems slow.  Then actual cryo-prep was begin 21 hours after death.  This is a far cry from the "Man down!  Cannot resuscitate!  Flush his fluids and pop him in the field cryo-chamber! Go, go, go!" of science fiction (I'm thinking Bujold here).

The panelists did have some things to say that suggested that some of the more recent corpsicles might not be complete writeoffs; I surmise that, someday, they may be usable as organ donors (paging Larry Niven).  The panelists, of course, are more optimistic.

Anyway, my forecast:

Sometime within the next century, there will be a successful cryo-revival.  But!  The sequence of events will be something like this.

A live rat will be prepared, frozen for a couple of weeks, and then revived.  It will have severe brain damage, and won't live long after revival, but, for a brief time, it will technically be a live defrosted cryo-rat.

Within a few years after the above success, a rat will be frozen and revived with its little rat brain intact and its other organs in decent condition.

All corpsicles with frozen-on dates prior to the above will have no chance of revival to functional-human state.  Until that first rat is revived, we don't know the right process for freezing them.  The ones frozen with wrong processes will be non-revivable.

Still... if people want to spend their money and their afterlives on this sort of thing, who am I to stop them?  If you're frozen with current technology, your chance of revival is laughable; if you're burned or buried (or dumped or eaten), it's zero, barring reincarnation or Riverworld.  Just remember to take out a good death-insurance policy, so if you're brought back to life you'll have money.

Insignia?

The Transylvanian Air Force, right?

(Photographing these dark-but-shiny things is tricky.  I guess I should have set up a light or two well off the camera axis.)

Shininess and stupidity

[NOTE: updated; see end of post.]

I tried to fill up the rest of the iPod.

Now, it might be that I did something somehow wrong, but... I transferred over a bunch more playlists and the associated MP3 files.

The last few CDs' worth came up with vacant playlists; when I tried to play one, the critter needed resetting.

Could this be some arbitrary limit?

The outside of the box said "500 songs."  Is the directory structure limited to 500 entries, as was the case (with maybe a different arbitrary number) with my mother's 1GB Muvo?

If so, there's great stupidity encapsulated withing the shiny shell of the iPod.  1.8GB @ 128kb/sec = 121ksec = 33.5 hours.  A 500 track limit gives an average of 4 minutes/track.  I guess that's not a totally unreasonable expectation, but not every track is a song, and some songs are short, and... well, I just don't like this sort of arbitrary limit.

If I'd paid for the thing, I'd be annoyed.

Update: Hokay, so now I see what the problem is, at least to a first approximation.

I'm not running up against some arbitrary limit of the iPod.

What's happening is that, when I copy over files that don't have any tags, they're ending up with null names, or something.  I don't know how much of this is the iPod, and how much is Amarok.  I'm also not sure how I ended up with rather a lot of tagless MP3 files; presumably I changed a setting in Grip somewhere along the line.

The playlists I copied over earlier were for files captured with one setting (which filled in the tags), and the ones I copied over later were, apparently, for files captured with the other setting.

Guess I need an MP3 file tag repair utility.  And to change my Grip settings.

Further update: filling in the ID3 artist/album/title tags does indeed solve the problem.  I managed to stuff the critter to within 14 megs of being full, and, thanks to the twin miracles of playlists and a multi-line screen, it's still usable.  Battery life looks promising, too.

Unsecured embedded systems

Some organizations just have no concept of security....

First there was the report that a (Diebold?) voting machine could be reprogrammed by opening the cover (with a common office-desk key) and inserting a memory card.  Now there's this report that a common Dutch voting machine can likewise be easily opened, allowing replacement of the PROMs.

You'd think the election authorities would take measures to discourage this.  Like, putting a serial-numbered seal on each machine, to that any tampering could at least be detected.  That works regardless of the machine design, unless the machine has some really stupid feature like firmware updates over Bluetooth.

And, on the machine-design front... these days, the program really ought to live in soldered-down flash, with updates through some sort of port that's inaccessible when the lid is closed... and with the reflashing software insisting on a valid digital signature before it'll proceed, and the boot code rechecking the signature (which doubles as a checksum) before running the application.

I carefully did not include remote firmware update capability in the embedded controller a client just shipped, even though that would have been a useful feature, because there wasn't time to secure it.  To upgrade the firmware, you need physical access to the controller board, which means tearing the whole subsystem apart.  Inconvenient, yes, but the customer doesn't have to worry that someone will sneak in and install unauthorized firmware with some sort of logic bomb lurking in it.

The second release of AGROS (I'm not up to the first public release yet; stay tuned) will support remote updates, with security features.  That's for eventually.  Some things, if they can't be done securely, are best not done at all.

Schwag!

Just back from the ARM Developers' Conference, with a bag of things.

Lots of potentially-useful information, of course - after all, that's the excuse for spending a few hours there.

Also a small beachball, a coupla T-shirts, and... a 2GB iPod Nano!  The first booth on the way in was Synopsys, and they were having a drawing, and not many people were there yet at 11:00, and...!

So now I'm learning about iPodim.  Linux recognizes it as /dev/sdb; there's a VFAT partition at /dev/sdb2 - for some reason, the first 24M and the last 1.5G show as unallocated; I'm guessing that I can expand partition 2 to fill the available space, but probably want to do some research before arbitrarily firing up gparted.  For loading music onto it, gtkpod works, and uses playlists - yay!  Got Die Walküre loaded on it, and the tracks are playing in sequence (this not being an avant-garde production, the orchestra is not playing in sequins).

Let's see, now: 2G is plenty for the entire Ring, right?

eric@verence:/media/mp3/richard_wagner$ du -s -h
808M    .

Yup.  Toss in a few other operas, too... this oughta keep my ears busy.  Now I need a good pair of folding headphones.

The TI booth had a drawing for something more useful - a TMS470 kickstart kit.  I didn't win that one, but they were offering discounts that bring the price down to something I can justify spending (and that, after all, is a business item).

Update: Amarok seems to be a pretty good program for managing an iPod under Linux.  It shows 1.8G usable, so I guess the funny partitioning is just Apple being what Apple is.  Anyway, I put the entire Ring on it, and have a gig left for, lessee, Aida, Rigoletto, Die Zauberflöte, all of Beethoven's symphonies, ...!

And I got plenty CDs, so who needs iZunes anyway?

Paging Jules Verne...

Slashdot links to this story about a scheme to flang space hardware from surface to orbit using a sort of macroscopic-particle cyclotron thingy to accelerate it gently to Mach 23 before releasing it skyward.

(Wouldn't actually put it in any kind of useful orbit, of course; there's still the need for some kind of orbital insertion burn, post-boost.  Remember, the orbit always passes through the point where the last force was applied.)

Is it just me, or is this a glorified version of Verne's Moon cannon?  And don't I recall seeing at least one detailed analysis of why, exactly, trying to impart such high velocities anywere near sea level is a Bad Idea?  Like, air looking very much like a brick wall when you're going that fast?

(Come to think of it, though, that analysis might have been by Arthur C. Clarke, who's been gloriously wrong about other things.)

Warning labels

Great commentary on a stupid warning-label trick here.  (Via the Volokh Conspiracy, where else?)

Short form: falling out of an upper bunk can hurt.  Don't do it.  Instead, go read the long form at the aforelinked article.

A fascinatingly goofy thought

My mother just got herself one of those newfangled bionic knees.  (She's not running 100 MPH yet.  In fact, she only got one bionic knee, which conjures thoughts of Attack of the Killer Tomatoes.)

Anyway, this got me to pondering materials, and the reasons titanium has been used for prosthetic structures, and....

And, hey.  Has anyone contemplated making aircraft out of bone?

Why not?  Train a chicken-bone tissue culture to grow to the specified form - maybe on the inside of an aircraft-shaped mold? - then dry it out.  You get a rigid structure, smooth and hard on the outside, hollow on the inside, and porous at the boundary.  You'd have to persuade it to grow around, and bond to, fittings for things like engine mounts, but....

OK, so it's a totally goofy way to make low-radar-signature airframes, and we probably don't have the biotech for it.  Yet.

But just you watch.  In 2014, we'll see the F-27 stealth fighter, made with organic technology.

And then the Shadows will have won.

Yeesh.

First Signetics changes its name to Philips Semiconductor, and then to Lowe's Hardware....

Now it's "NXP Semiconductors."  (I wouldn't normally know that this early in the morning, but Digi-Key sent me an announcement of the change in the wee hours.)

NXP?  What the heck kind of name is that?  Doesn't "NX" usually stand for "No Such," as in ENXIO?  Does the new name mean "No Such Philips," or "Philips ain't here, man?"

Ah, well.  The good news is that the new web site is much faster, once you get past the fool flash intro.  And it looks like the LPC2100 family survived the name change (LPC2148, check; LPC2134, check; OK, those are the ones I'm currently using).

Update: Grrrr.  Their web designers have gotten cute, and datasheets and app notes now arrive as ZIP bundles selected from pop-up checklists triggered by onmouseover events, instead of just having individual links to the PDFs.  (I'll have to try with Jawascript turned off sometime, to see if the links are still lurking behind the new tomfoolery.)

Come visit the kernel

Gumbyware was off the air for quite a while early this morning....

There'd been FreeBSD security alerts accumulating for a while, and since the ones the last few days had been related to OpenSSL and OpenSSH, which are exposed to the cruel world, I figure I'd better do something about them.

So:

# cvsup -g -L 1 -P - -z /etc/cvsupfile
# make buildworld
# make installworld
# reboot

And all should be well, right?

Er... everything was fine, right up until the startup scripts reached NFS, at which point there was a kernel panic, core dump, and emergency reboot... which failed in exactly the same way.  (After a couple of these failures, at least I'd figured out that it was happening during NFS startup.)

So, futz around with booting into single-user mode (all the other nominally-safer-than-normal modes were still trying to start NFS), remounting / to allow writing... then edit /etc/rc.conf to change the NFS options from YES to NO.  Reboot.

OK.  Now it's booted OK, and outside services (web, mail, subversion, and so on) are running.  I can't use my workstation, though, because it's still waiting for /home to reappear on the server... via NFS, of course.

Take a wild guess that there are updates to the NFS subsystem which are now out of sync with the kernel (I suppose that can happen when you upgrade all the services to 6.2 while leaving the kernel at 6.0).  So,

# make buildkernel KERNCONF=VETINARI
# make installkernel KERNCONF=VETINARI
# reboot

Yay!  It comes up without crashing, and I can then re-enable and start NFS, and it doesn't freak out.  Do one last reboot, with NFS turned on in rc.conf, just to make sure.  Success!

And now, while I'm thinking about it, methinks it wise to burn a 6.2 rescue CD.  Just in case.

Update: Burning a CD this week is more trouble than I'd expected.  The recent Debian upgrade replaced cdrecord with the splinter project wodim (I guess I'd heard something about a falling-out between Schilly and the Debian team, a few months back), and it doesn't seem to be well synchronized with k3b yet.  Invoking wodim from the command line with minimal options seems to work fine, for an existing ISO image, but k3b sometimes works and sometimes doesn't, in strange and wondrous ways (exiting and restarting k3b seems to be the cure).  Also, xcdroast complains that it can't find /usr/bin/cdrecord (which is now a symlink to wodim), and aborts.  There seems to be some issue with growisofs, too; it wrote a little and then died, the couple of times I tried burning a data DVD from k3b (I ended up dividing the data across 2 CDs instead).

Ah, well... time to check for updates.  Life running -unstable is so exciting, it's almost like Microsoft!  Well, except for the viruses.