[erlang-questions] State of the Union: FFI

[Serge Aleynikov]

That would be good.

I also would like to express keen interest in pushing this EEP forward 
in the queue so that the OTP team would consider its inclusion in the 
distribution.  This EEP has been sitting out there for a while and if 
there is a reason why it's not being processed, please let us know.

Though driver-based C interop API is fast, it does incur around 20-30us 
overhead for a function call in the current implementation (though I 
last benchmarked it in R11B-5), whereas an FFI-based approach can be 
made more optimal (at the cost of potential loss of robustness, yet 
often that is an acceptable sacrifice for the sake of performance gain 
when bridging Erlang to other C/C++ libraries).

Serge

Zvi wrote:
> I propose to not wait for standardization, and port it ourselves to R13A.
> It's a crucial feature for my application. Even if later I will need to port
> all my code to linked-in driver, ffi allows me to work right now and rapid
> prototyping just can't be achieved with linked-in drivers.
> There are several people interested.
> I even may pay some symbolic sum to speed things up.
> 
> BR,
> Zvi
> 
> 
> Franco Milicchio-4 wrote:
>> Dear all,
>>
>> Is there someone that knowns the inner working of FFI standardization?  
>> I am trying to use it, but it's really not a piece of cake when I must  
>> remain with older versions of Erlang/OTP.
>>
>> By the way, is there someone I can talk with regarding FFI? (in  
>> production, of course!)
>>
>> Thanks!
>>
>>
>> PS. I know this is a recurring topic, but I didn't find any post newer  
>> than six month, so... this is just a reminder :)
>>
>> -- 
>> Franco Milicchio <fmilicchio@REDACTED>
>>
>> DIA - Dept. of Computer Science and Engineering
>> University Roma Tre
>> http://plm.dia.uniroma3.it/milicchio/
>>
>>
>>  
>> _______________________________________________
>> erlang-questions mailing list
>> erlang-questions@REDACTED
>> http://www.erlang.org/mailman/listinfo/erlang-questions
>>
>