On Fri, Nov 6, 2015 at 6:15 PM, Dmitry Gutov <dgutov@yandex.ru> wrote:
> That sounds fine to me in terms of design, but it might add some 
> performance overhead. So some testing is needed.

Good! A revised patch is attached.

Benchmarks follow below, with the same setup as last time.

Before the patch:

(benchmark-run 100 (json-read-from-string huge-json))
     ⇒ (16.84457266 1007 4.886441912999999)

After the patch:

(benchmark-run 100 (json-read-from-string huge-json))
     ⇒ (16.905379125000003 1007 4.722544520000007)

-- Simen