On 29/03/2008, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> I intentionally didn't touch xml.c, nor anyplace that is not dealing
> in text, even if it happens to be binary-compatible with text.
>
Hmm, okay. My original submission did include a few such changes; for
example, in xml_in and xml_out_internal I saw that the conversion
between xmltype and cstring was identical to the text conversion, so I
went ahead and used the text functions. Feedback upthread suggested
that it was okay to go ahead with casting in identical cases. [1]
I saw that these changes made it into the commit, so I assumed that it
was the right call.
If we don't want to meddle with xmltype/bytea/VarChar at all, we'll
have to revert those changes, and I'll have to seriously scale back
the cleanup patch I was about to post.
Regards,
BJ
[1] http://archives.postgresql.org/pgsql-hackers/2007-09/msg01094.php