A collegue of mine is having the same problem. It appears on
machines running RedHat Linux 4.2 & 5.0 (It works fine on SGI, an old
Slackworks (?) Linux,...)
It doesn't seem to care which version of l2h or perl is running (?)
He narrowed it down to the tell & seek in
sub slurp_input_and_partition_and_pre_process
(are those still needed?)
_A_ workaround, but likely not the `Correct' solution was to put a
close SINPUT;
just after
unless ($pid = fork) {
in sub write_string_out
Interestingly, the comments for the tell & seek statements claim that they are
there to get around some sort of problem w/ forking.
If we comment out the tell & seek, but include the close, it still works fine.
If we comment out tell & seek AND the extra close, it produces 3,
count 'em THREE, copies of the section!
More bizarre: I tested this on my SGI, which had worked fine.
Commenting tell & seek, but including the close produces 3 copies of section.
Having tell & seek AND the close works fine (again).
Apparently SGI (and others?) need the tell/seek `solution', and RedHat needs
the close `solution'... or is something else really going on?
Thanks
--
--
[EMAIL PROTECTED]
http://math.nist.gov/~BMiller/
