Re: Another swing at JSON - Mailing list pgsql-hackers

From Bernd Helmle
Subject Re: Another swing at JSON
Date
Msg-id EC307F0B4E0783FF739E0162@apophis.local
Whole thread Raw
In response to Re: Another swing at JSON  (Joseph Adams <joeyadams3.14159@gmail.com>)
Responses Re: Another swing at JSON
List pgsql-hackers

--On 29. März 2011 21:15:11 -0400 Joseph Adams <joeyadams3.14159@gmail.com>
wrote:

> Thanks.  I applied a minor variation of this trick to the JSON module,
> so now it builds/installs/tests cleanly on both REL8_4_0 and HEAD
> (though it won't work if you copy contrib/json into a pre-9.1
> PostgreSQL source directory and type `make` without USE_PGXS=1).
>
> I also went ahead and renamed uninstall_json.sql to
> json--uninstall--0.1.sql (again, it's for pre-9.1 users) and removed
> unnecessary trailing spaces.
>
>> Anything going into the PostgreSQL code base will be for 9.2, so
>> anything else would be a separate (if somewhat related) project.  I
>> suspect the code will be a good deal cleaner if you do just the 9.2+
>> version and see who wants it back-patched, if anyone does :)
>
> It's a trivial matter to remove backward compatibility from
> contrib/json, if anybody wants me to do it.  I can just remove
> compat.[ch], */init-pre9.1.* , remove the PREFIX_PGVER trick from the
> Makefile, remove a few lines in the source code, and maintain the
> backported json module elsewhere.  It's just a matter of whether or
> not explicit backward-compatibility is desirable in modules shipped
> with releases.

I started looking into this. A very minor adjusted patch to filelist.sgml was
required to apply the patch cleanly to current -HEAD (attached).

After reading Joseph's comment upthread, I don't see any consensus wether the
existing pre-9.1 support is required or even desired. Maybe i missed it, but do
we really expect an extension (or contrib module) to be backwards compatible to
earlier major releases, when shipped in contrib/ ?

--
Thanks

    Bernd


Attachment

pgsql-hackers by date:

Previous
From: Bruce Momjian
Date:
Subject: Re: procpid?
Next
From: Tom Lane
Date:
Subject: Re: Boolean operators without commutators vs. ALL/ANY