Re: Initial review of xslt with no limits patch

On 08/06/2010 02:29 AM, Pavel Stehule wrote:
> 2010/8/6 David Fetter<david(at)fetter(dot)org>:
>> On Fri, Aug 06, 2010 at 05:57:37AM +0200, Pavel Stehule wrote:
>>> 2010/8/6 Andrew Dunstan<andrew(at)dunslane(dot)net>:
>>>> On 08/05/2010 06:56 PM, Mike Fowler wrote:
>>>>> SELECT
>>>>> xslt_process('<employee><name>cim</name><age>30</age><pay>400</pay></employee>'::text,
>>>>> $$<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
>>>>> version="1.0">
>>>>> <xsl:output method="xml" omit-xml-declaration="yes" indent="yes"/>
>>>>>
>>>> [snip]
>>>>> </xsl:stylesheet>$$::text, 'n1=v1,n2=v2,n3=v3,n4=v4,n5=v5'::text)
>>>> I haven't been paying attention to this, so sorry if this has been discussed
>>>> before, but it just caught my eye. Why are we passing these params as a
>>>> comma-separated list rather than as an array or as a variadic list of
>>>> params? This looks rather ugly. What if you want to have a param that
>>>> includes a comma?
>>> There is probably problem in pairs - label = value. Can be nice, if we
>>> can use a variadic functions for this, but I am afraid, ...
>>>
>>> using a variadic function isn't too much nice now
>>>
>>> some xslt_process(xmlsrc, 'n1=v1','n2=v2','n3=v3'
>> This sounds like the perfect case for pulling hstore into core code. :)
> I afraid so integration of hstore can break and block work on real
> hash support. I would to have hash tables in core, but with usual
> features and usual syntax - like Perl or PHP
>

Can we just keep this discussion within reasonable bounds? The issue is
not hstore or other hashes, but how to do the param list for xslt sanely
given what we have now. A variadic list will be much nicer than what is
currently proposed.

cheers

andrew