This attached patch greatly enhances DBMirror.pl performance.
DBMirror.pl was known to have problems when replicating
bytea columns whose data was binary bytes (non printable).
(E.g. tiff, pdf, jpeg, bzip2, etc...)
Minutes, (or hours) for 500kb columns were not unusual.
There has even been an effort by Peter Wilson
(<petew ( at ) yellowhawk ( dot ) co ( dot ) uk>)
to write a C alternative,
which tries to overcome these problems.
However i think another C program is not really needed
at this point.
This patch changes the way extractData parses the data field.
Please have a look at the patch, if you see any potential problems.
Thank you.
P.S.
I also emailed the -sql,-general,-pgreplication lists
with a relevant message.
Steven Singer (the original author) is not reachable.
However i think that some users might already suffer from
poor DBMirror.pl performance.
--
-Achilleus
