Re: patch: function xmltable

From: Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com>
To: Alvaro Herrera <alvherre(at)2ndquadrant(dot)com>
Cc: Craig Ringer <craig(at)2ndquadrant(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: patch: function xmltable
Date: 2016-11-24 05:04:23
Message-ID: CAFj8pRCPm2Y76NtWM1z1tFyWttyp482cAgF0+PH7+zZs34NysQ@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox
Thread:
Lists: pgsql-hackers

2016-11-24 0:29 GMT+01:00 Alvaro Herrera <alvherre(at)2ndquadrant(dot)com>:

> Here's another version. Not there yet: need to move back the function
> to create the tupdesc, as discussed. Not clear what's the best place,
> however. I modified the grammar a bit (added the missing comma, removed
> PATH as an unreserved keyword and just used IDENT, removed the "Opt"
> version for column options), and reworked the comments in the transform
> phase (I tweaked the code here and there mostly to move things to nicer
> places, but it's pretty much the same code).
>
> In the new xpath_parser.c file I think we should tidy things up a bit.
> First, it needs more commentary on what the entry function actually
> does, in detail. Also, IMO that function should be at the top of the
> file, not at the bottom, followed by all its helpers. I would like some
> more clarity on the provenance of all this code, just to assess the
> probability of bugs; mostly as it's completely undocumented.
>
> I don't like the docs either. I think we should have a complete
> reference to the syntax, followed by examples, rather than letting the
> examples drive the whole thing. I fixed the synopsis so that it's not
> one very long line.
>
> If you use "PATH '/'" for a column, you get the text for all the entries
> in the whole XML, rather than the text for the particular row being
> processed. Isn't that rather weird, or to put it differently, completely
> wrong? I didn't find a way to obtain the whole XML row when you have
> the COLUMNS option (which is what I was hoping for with the "PATH '/'").
>

This is a libxml2 behave

Postprocessing only check result and try to push the result to expected
types.

Regards

Pavel

>
> --
> Álvaro Herrera https://www.2ndQuadrant.com/
> PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
>

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Karl O. Pinc 2016-11-24 05:08:18 Re: Patch to implement pg_current_logfile() function
Previous Message Tsunakawa, Takayuki 2016-11-24 04:59:25 Re: Re: [bug fix] Cascading standby cannot catch up and get stuck emitting the same message repeatedly