Re: using tablewriter with binary data

> Oops, you're right.  I've fixed this in revision [1265].  Thanks for a
> thorough report.

Thanks! However, an issue remains.

In this case I am writing to a table with one column being a bytea
(rather than a varchar). The r1265 change unfortunately does not solve
the problem, since there seem to be two levels of unescaping involved.

Looking at the output of pg_dump, binary values have their backslash
escaped aswell (so \\000 instead of \000). I will admit that I find
the PostgreSQL's escaping to be quite strange and I never understood
why there is not simply a single escaping scheme regardless of
binary/non-binary data. In this particular case I suppose one level of
unescaping is being applied by the COPY receiver, and the second level
further into the backend where newlines and such need not be escaped,
but binary data does.

In order to make the table writer work in the case of inserting binary
data into a byteaa column I had to further make this modification:

-      R += '\\';
+      R += "\\\\";

Note that one specifically should not do this in the escaped character
case (i.e., newlines are \n, not \\n), which is consistent with the
above mentioned theory on escaping.

This solves the problem in the case of inserting binary data into a
bytea column, and still seems to work in the case of varchar:s (i.e.,