Some sort of size limitation about jiffy HOT 3 CLOSED

I'm not inclined to have such a limitation. Seeing as term_to_binary doesn't have anything like that I'm not inclined to add one here. Although, a related thing that I haven't gotten around to is to make sure and bump the reduction count. It might be a good idea to figure out something there. If you do look at that please to open a new ticket for it.

from jiffy.

It's not just about bumping reduction count, it's also about blocking the scheduler (AFAIK scheduler can't reschedule current reduction if it is in NIF). Actually I don't have this issue right now but I can suffer from it in a not so distant future, so I'm wondering what do you think about this.

from jiffy.

Right, my point is that the same exact thing affects term_to_binary/1 and binary_to_term/1. Most people are concerned about NIFs because of the stern warning in the docs, but if we compare to existing BIFs then there's no issue. If the size of JSON inputs becomes such an issue that it does have negative effects then the only real answer is to use a pure Erlang parser and take the performance hit.

from jiffy.