[erlang-questions] suggestion: NIF_D

Ulf Wiger <>
Tue Apr 20 19:38:43 CEST 2010

Paul Davis wrote:
> In the case of "lets take a look at writing NIF's with D" I was trying
> to say that the scope of a NIF is somewhat limited by the fact that it
> must return quickly or risk locking the entire Erlang VM. Due to this,
> the LOC in the NIF's I've seen is quite small (1K is a big NIF). With
> such a small foot print I wouldn't expect the benefits of using a
> language like D to outweigh the costs of depending on its build chain.

Well, the requirement to return quickly doesn't necessarily mean that
the code size will be small. The line count of beam/erl_db* (the
ETS BIFs) sums up to 11K. :)

I agree that the NIFs to-date have mainly been small. I fully expect
to see some large ones in the near future.

> In the case of "lets take a look at providing a platform to run
> arbitrary D in Erlang" I reckon that could be pretty awesome but I am
> unsure whether it would be enough awesome to depend on whatever
> build/runtime dependencies it required.

This is a real concern, of course. Not having used D, I have no clear
intuition for the pain level this would cause, but I can definitely
imagine that it could be significant.

Ulf W
Ulf Wiger
CTO, Erlang Solutions Ltd, formerly Erlang Training & Consulting Ltd



Since January 1st 2010 Erlang Training and Consulting Ltd. has become ERLANG SOLUTIONS LTD.


More information about the erlang-questions mailing list