Thank you dfs and shotgunefx, two great ideas.
The big problem was browser timeout, which had not solved
itself by just printing dots. I do dots a lot, maybe spaces
that don't display would be good. For some reason I
remember thinking the browser was timing out even with the
dots being printed, but obviously the process must have
just taken too long.
I had looked at IO::Tee (output to multiple handles) but
I guess the answer is really just a fork as shotgunefx
suggested. The Perl program directs the spidering and
indexing of many websites into separate databases, so
I would fork a child for each site, then have the child
return a success code on finishing its work. I suppose
this would even run more quickly with many children in
parallel but I think I'll keep it serial for now to keep
the log file readable.
Thanks. I guess I was a little *too* lazy a perl programmer.