[plt-scheme] Odd CTRL-C behavior

Hi all,

This is almost certainly a Matthew-question, but perhaps someone else
has some insight.

My co-conspirator Christian Jacobsen has been exploring exception
handlers in our stand-alone linker. In particular, he was looking to
catch CTRL-C, and provide a slightly less verbose and
end-user-unfriendly stack-trace. Or, if you prefer, he wanted to
provide a simple error message (or perhaps none at all, even).

What he discovered was that depending on when he hit CTRL-C, he could
get different behaviors. Christian assured me this was a Very
Scientific Process, and that it was Ultimately Repeatable if he was
Held Excessively Accountable.

His theory is that there is a period of time before which our compiled
code has not yet been executed, and we might get any number of errors
as the runtime support environment for a stand-alone executable comes
to life. Does this sound reasonable/likely?

Given his experimentation, and the nature of the inquiry, this does
not strike me as a behavior we can get away from. Does anything
obviously come to mind how we can protect end users from odd error
behavior on break before our exception handler is registered? (Mind
you, I don't actually think we have so many users, but that's besides
the point. I thought I'd ask anyway.)

Many thanks,
Matt