Re: nested hstore patch

On 01/13/2014 11:03 AM, Alvaro Herrera wrote:
> Andrew Dunstan wrote:
>> On 01/13/2014 03:25 AM, Erik Rijkers wrote:
>>
>>> There are errors in the example expressions in "Table F-6. hstore Operators".
>>>
>>> Attached is a cumulative doc-patch (which includes the changes I sent earlier) which fixes these.
>>>
>>> I also attach an test perl program that shows the (small) differences in output between what's in that doc table
andwhat
 
>>> one actually gets. (I found these too insignificant to change but perhaps you have a different opinion.)
>>
>> A new version of the patch is attached. It includes all of Erik's
>> docs fixes and a small fix by Alexander Korotkov for hstore hash
>> ops.
> Interestingly, this also include transaction_commit event triggers.

Oh, wow, really? git really did something horrible, or I did 
inadvertently. This is what comes from using the same directory for 
multiple development lines :-(


Will fix

>
> There are also a few PANIC elogs, probably not what's intended.


Oleg, Teodor, please address.

>
> (I was just giving this a quick skim to see if there's support to build
> JSON objects incrementally from C source, i.e. not have to call
> functions using the fmgr interface.  Apparently that's not the case, but
> if I'm wrong please let me know.)


Erm, maybe you need the other json patch: 
<http://www.postgresql.org/message-id/52C76B33.1050808@dunslane.net>

If we need to adjust some of that a bit to make it more friendly for 
internal use I'm happy to try to do that. Unfortunately, I don't think 
that's terribly easy for VARIADIC "any" functions like these.

cheers

andrew