Impact of native compilation

[Marc Feeley]

> What's the word on ETOS' availability? Last year I mailed in asking
> for the sources to the lastest version, but didn't hear back. The
> website only has 2.3 in binary, and 1.4 as source.

ETOS has not progressed since version 2.3 and it is unlikely that my
group will ever finish the implementation completely (i.e. conformance
to the Erlang spec).  Remember that ETOS is a research project whose
aim is to show that it is possible to quickly implement a compiler
from Erlang to Scheme that achieves good performance.  I think we
attained that goal with ETOS even though not all of Erlang's libraries
are implemented.  Implementing good Erlang librairies takes a lot of
time, and there is unfortunately very little "academic reward".

Given that Hipe has reached (and most likely by now exceeded) the
performance of ETOS, and Hipe implements all of Erlang, there is
little real need to continue the development of ETOS.

Note that I am still interested in Erlang.  I think there are many
lessons to learn from it.  Some of the features are really important
and relatively unique (distribution, communication, dynamic code
loading).  There are also some problems with Erlang's syntax and
semantics (exceptions, records, types, macros, distributed garbage
collection, etc).  One of my pet projects is a redesign of an
Erlang-like language "from scratch".  So stay tuned... but then again
don't hold your breath!

Marc