[erlang-questions] Teaching Erlang as part of a paper -- advice sought

Richard O'Keefe <>
Tue Feb 9 00:12:30 CET 2010


On Feb 8, 2010, at 10:33 PM, Ulf Wiger wrote:
> Kostis has alreay mentioned that data races are possible in Erlang.

I am dusting off my old 'pid_name' proposal from the 90s and turning
it into an EEP.  I'd like to avoid races in the core language.
>
> One possible benefit of using your approach is that it can highlight
> that race conditions come in different forms. Some are part of the
> problem domain (e.g. handshakes in peer-to-peer networks, credit
> card transaction processing, session management in messaging systems
> in general, etc.) while others are accidental - a consequence of the
> chosen implementation technique. I would think that reasoning about
> the high-level races provides an ample foundation for talking about
> low-level data races.

I was proposing to have a couple of lectures about modelling systems
in some sort of process calculus.  Because I know NuSMV, that's what
I was planning to use.  I've always been very impressed by the work
Thomas Arts and his colleagues did, but the latest version of EVT I
can find is 2001, and since then, not only have Linux and Solaris
and Java changed, so has Erlang changed.

Ulf, can you recommend anything I should look at to understand the
issues in "(e.g. handshakes in peer-to-peer networks, credit
> card transaction processing, session management in messaging systems
> in general, etc.)"?


> BTW, which mutual exclusion techniques
> can not be illustrated in Erlang?

None that I know of.  At least, no _high level_ ones.  If you want
to demonstrate how, e.g., spin-locks _work_ at a low level, that's
another matter.



More information about the erlang-questions mailing list