Sun Dec 15 23:56:23 CET 2013
On Sun, Dec 15, 2013 at 10:39 PM, Serge Aleynikov <serge@REDACTED> wrote:
> Since websockets nowadays support binary data, I implemented binary Erlang
> External Term serialization, so that term_to_binary(Term) can be decoded
> I'd like to share this example that illustrates the experiment:
Are you going to use it practically? If yes, my rant below.
I remember Joe's opinion (from Stockholm EUC keynote last year?).
Imprecise quote from memory: "if everyone could speak Swedish as well
as native language, we all could talk to each other". No interpreters:
just convert your native sentence to Swedish, speak out, the recipient
converts Swedish to native, understands, replies ... all is good.
The whole idea (and your project) pushes towards "Swedish=Erlang Term
Format". This is all nice, but has a big IF: since this concerns
communication between heterogenous computer systems, how are you going
to convince enterprisey mainstream guys to talk to you via BERT? If
all they know is XML (SOAP!), JSON (front-end) or HDF (CERN guys) or
stuff like that.
I believe in an approach piqi is trying to do: an interpreter which
can speak much of the popular languages *and* your mother tongue. It's
different, as it forces a schema (unlike BERT). However, for data
exchange we need a schema anyway.
Not meaning to offend/criticize. What you did is nice. Your project
just reminded me an unanswered question.
More information about the erlang-questions