Re: __FUNCTION__ and debug.

On Sunday 27 October 2002 05.02, Adrian Chadd wrote:

> Right. Hm. Are there any reasons why we'd want to _not_ have a
> single trace call map to a single debug line? I can't think of any,
> but ISTR some magical stuff in some file which used it for
> formatting. I could be wrong.
>
> Should we call it debug_trace() or something a little more
> descriptive? trace() might sound like it has stack traces with it
> or something. :)

I am all for the principle of automatically emitting which function
the debug trace is from, but agree with Adrian that the trace() name
is perhaps a bit ambigious.

Why not simply let it be debug() and gradually clean up the debug
statements who manually emits a function name to not do so..?

What I don't quite get is a minor syntactic thing in how you convince
the compiler to add __FUNCTION__ to the format string with the
proposed syntax, but I guess it is all possible with some OO
trickery..

It might need a sister who writes raw information to the debug log,
i.e.

debug_raw(section, level)(format, ...)

which sends the data (if below the fence) as is to the debug log, not
even a timestamp I think.

Regards
Henrik
Received on Sun Oct 27 2002 - 10:09:27 MST