Re: xpath changes in the recent back branches

From: Mike Rylander <mrylander(at)gmail(dot)com>
To: Marko Tiikkaja <marko(at)joh(dot)to>
Cc: Pgsql Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: xpath changes in the recent back branches
Date: 2015-03-04 16:48:32
Message-ID: CAO8ar=n1pGwe=QSMVG88604qz07NBXkdH9Fi3WUwjU7ay6__ag@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Thu, Feb 19, 2015 at 5:53 AM, Marko Tiikkaja <marko(at)joh(dot)to> wrote:

> Hi,
>
> Commit 79af9a1d2668c9edc8171f03c39e7fed571eeb98 changed xpath handling
> with regard to namespaces, and it seems to be fixing an actual issue.
> However, it was also backpatched to all branches despite it breaking for
> example code like this:
>
> do $$
> declare
> _x xml;
> begin
> _x := (xpath('/x:Foo/x:Bar', xml '<Foo xmlns="teh:urn"><Bar><Baz>1</Baz><Bat>2</Bat></Bar></Foo>',
> array[['x','teh:urn']]))[1];
> raise notice '%', xpath('/Bar/Baz/text()', _x);
> raise notice '%', xpath('/Bar/Bat/text()', _x);
> end
> $$;
>
> The problem is that there's no way to write the code like this in such a
> way that it would work on both versions. If I add the namespace, it's
> broken on 9.1.14. Without it it's broken on 9.1.15.
>
> I'm now thinking of adding a workaround which strips namespaces, but that
> doesn't seem to be easy to do, even with PL/Perl. Is there a better
> workaround here that I'm not seeing?
>
>
FWIW, I've been working around the bug fixed in that commit for ages by
spelling my xpath like this:

xpath('/*[local-name()="Bar"]/*[local-name()="Baz"]/text()', data)

I've modularized my XML handling functions so the source of 'data' is
immaterial -- maybe it's a full document, maybe it's a fragment from a
previous xpath() call -- and the referenced commit is going to make correct
XPATH much more sane, readable, and maintainable. I, for one, welcome it
wholeheartedly.

HTH,

--Mike

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Paul Ramsey 2015-03-04 16:58:33 Re: compress method for spgist - 2
Previous Message Jeff Janes 2015-03-04 16:41:23 Re: Comparing primary/HS standby in tests