Re: jsonb access operators inefficiency

Andrew Dunstan <andrew(at)dunslane(dot)net> writes:
> If we're going to construct varlena objects inside a StringInfo, maybe
> we need a proper API for it. Isn't there a danger that data member of
> the StringInfo won't be properly aligned to allow us to do this? In any
> case, we should get most of the benefit of your patch without this
> "optimization".

As noted, the data buffer is maxaligned; but nonetheless I agree that
this is a serious stylistic abuse, and it's not buying much compared
to the rest of the patch. I'd stick with the cstring_to_text_with_len
coding.

At the other end of the process, why are we using PG_GETARG_TEXT_P
and not PG_GETARG_TEXT_PP to avoid a "detoast" cycle on short-header
inputs? The function body is using VARDATA_ANY/VARSIZE_ANY_EXHDR,
so it's already prepared for unaligned input.

regards, tom lane