Re: Where does g++ keep temporary template source

[brian]

Yes, what I am describing is impossible. Except, it really is
happening. Which is why it is so perplexing.

Yes, ddd is my next step (will have to wait for Monday). But in case
there was something that g++ was leaving around that I didn't know
about, I thought I'd ask. Thanks for the confirmation.

I'm only guessing that the 2-parameter version is being called;
inspection and run-time checking of the code shows no other way that
the enumrated type data member could be set to an invalid value (only 0
and 1 are defined). In hindsight, I will likely find this guess to be
incorrect because that's how most of my guesses in this case turn out.
However, it provides the best description of the problem that I can
think of at this time.

By the way, Brian Kernigan talks about how he finds effective debugging
being careful design and a few well-placed print statements. Those, and
run-time traces, keep me out of the debugger for the most part. (And
the robustness and maintainability of my code generally far exceeds
that of those who live in the debugger. I think it's the careful design
part more than the print statements, though!) However, this will be one
of those nstances when I will unleash ddd/gdb on a problem.

Thanks again for the insights.