Re: fixing PQsetvalue()

On Wed, Jul 20, 2011 at 10:28 PM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
> On Mon, Jul 18, 2011 at 6:38 AM, Pavel Golub <pavel(at)microolap(dot)com> wrote:
>> Hello, Merlin.
>>
>> I hope it's OK that I've added Andrew's patch to CommitFest:
>> https://commitfest.postgresql.org/action/patch_view?id=606
>>
>> I did this becuase beta3 already released, but nut nothig is done on
>> this bug.
>
> So I finally got around to taking a look at this patch, and I guess my
> basic feeling is that I like it.  The existing code is pretty weird
> and inconsistent: the logic in PQsetvalue() basically does the same
> thing as the logic in pqAddTuple(), but incompatibly and less
> efficiently.  Unifying them seems sensible, and the fix looks simple
> enough to back-patch.
>
> With respect to Tom's concern about boxing ourselves in, I guess it's
> hard for me to get worried about that.  I've heard no one suggest
> changing the internal representation libpq uses for result sets, and
> even if we did, presumably the new format would also need to support
> an "append a tuple" operation - or the very worst we could cause it to
> support that without much difficulty.
>
> So, +1 from me.

right -- thanks for that. For the record, I think a rework of the
libpq internal representation would be likely to happen concurrently
with a rework of the API -- for example to better support streaming
data. PQsetvalue very well might prove to be a headache -- just too
hard to say. libpq strikes me as a 50 year plus marriage might:
fractious, full of mystery and regrets, but highly functional.

merlin