[tpm] can't deliver a talk tomorrow night

[arocker at vex.net]

Considering the very short notice, I think we did rather well last night,
(even if, as one of the presenters, I have a certain bias :-)* ).

As usual, in the process of trying to explain something, I probably learnt
more than the audience. (Thanks, Mike.)

A discussion started about the relative performance of a subustr function
and a regex for excising 8 bytes from a string. The difference turns out
to be a factor of about 20 (38.5 cpu-seconds vs 1.86) in favour of the
substr.

However, on a 3Ghz Core Duo box, Benchmark required 10 million iterations
to produce a reliable result, so I think it's safe to say it really
doesn't matter in most plausible circumstances.