Re: problem with precendence order in JSONB merge operator - Mailing list pgsql-hackers

From David G. Johnston
Subject Re: problem with precendence order in JSONB merge operator
Date
Msg-id CAKFQuwb6GXMWu95_fSbxjHi=qwqrcUCpqTG-Qr-KL_mAM6H4Mw@mail.gmail.com
Whole thread Raw
In response to Re: problem with precendence order in JSONB merge operator  (Peter Krauss <ppkrauss@gmail.com>)
List pgsql-hackers
Please don't top-post.

On Tuesday, March 22, 2016, Peter Krauss <ppkrauss@gmail.com> wrote:
Subjective notes to contextualize (try to explain on bad-English) my "precedence order" and JSONB visions:

JSON datatype is perfect as workaround, and for many simple and less exigent applications.
JSONB is the  "first class" datatype for user community, we expected years (!) for it ... Need some "first class" and friendly behaviour. 

In this context JSONB is not "any other" datatype, it is the bridge between relational data and flexible data...
It is the Holy Grail and the Rosetta Stone :-) 

I think JSONB operators need some more attention, in semantic and usability contexts.   If you want to add  some friendliness and orthogonality in JSONB operators, will be natural to see -> operator as a kind of object-oriented path operator... 
By other hand, of course, you can do the simplest to implement JSONB... But you do a lot (!), it was not easy to arrive here, and need only a little bit more to  reach perfection ;-)


You are welcome to supply a patch for this particular "little bit" - but I suspect you will find it quite disruptive to backward compatibility and general system internals if you attempt to do so.  But maybe not.

Any change you make in this area will effect every usage of that operator whether part of core or end-user defined.  We have baggage that limits our ability to be perfect.

So while I'll agree with your premise I don't see (really, don't care to look for) a way to make your desire a reality.  But you and smarter people than I are welcome to dive into the code and see if you can come up with something that works.

David J.

pgsql-hackers by date:

Previous
From: Craig Ringer
Date:
Subject: Re: Timeline following for logical slots
Next
From: Michael Paquier
Date:
Subject: Re: WAL logging problem in 9.4.3?