sed-3.02.80

[Jan van der Steen]

Dear sirs,

I have a peculiar problem with GNU sed.
While converting a HTML file the first
step is to join all lines using tr '\n' ' '.
As you can imagine this gives some pretty
long lines for some files. I have a problem
with a file exceeding 164000 bytes (this
number is quite acurate). The substitute command
which fails is:

    s#^.*<h2[^>]*#<h3#

When the input is smaller than 164000 bytes the
substitute command will succeed. When the input
is larger than 164000 bytes the command will fail.
The part before the "<h2" pattern is neglegible.
The part after is the bulk. If desired I can send
you the script and file to reproduce the problem.
I've tried both sed-3.02 and sed-3.02.80, both
on Linux as on Sun Solaris.

kind regards, Jan van der Steen