Re: XML binary I/O (was Re: tsearch refactorings)

Heikki Linnakangas wrote:
> Heikki Linnakangas wrote:
>> BTW, the encoding of the XML datatype looks pretty funky. xml_recv first
>> reads the xml string with pq_getmsgtext, which applies a client->server
>> conversion. Then the xml declaration is parsed, extracting the encoding
>> attribute. Then the string is converted again from that encoding (or
>> UTF-8 if none was specified) to server encoding. I don't understand how
>> it's supposed to work, but ISTM there's one conversion too much,
>
> And it's got an unfortunate typo in it as well: it calls "free(result)"
> instead of pfree. I think we need regression tests for the more complex
> send/recv functions...

According to the docs, xml_send is supposed to output the XML document
in client encoding, with that encoding specified in the xml declaration,
eg. ?<xml encoding="LATIN1"?>. xml_recv is supposed to ignore client
encoding, and parse the document according to the encoding specified in
the declaration.

Here's a patch that fixes the send/recv functions to work like the
manual says. It fixes the free/pfree typo as well.

Included is a new regression test for the send/recv functions as well
(unpatched CVS HEAD fails it BTW). It can be merged with the existing
xml test, but I didn't want to do it in the patch because it would've
involved moving the current xml test from sql/ to input/, and made the
patch longer to review.

--
Heikki Linnakangas
EnterpriseDB http://www.enterprisedb.com