[erlang-questions] Delivery of messages

[Matthias Lang]

On Thursday, February 23, P.H.Welch wrote:

> The most authoritative discussion I've found is question 10.10 at:
>
>   http://www.erlang.org/faq/academic.html

There are more in-depth discussions, mainly Hans Svensson's (and
collaborators) work a few years ago. The paper I remember is
'Programming Distributed Erlang Applications: Pitfalls and Recipes',
though there were others. Here's a link:

  http://www.cs.chalmers.se/~hanssv/doc/ew07-dist.pdf

which doesn't work anymore. Googling the title gets results, but they
look like they're all behind a paywall. Great.

>From memory, the paper demonstrated that at least some of the message
sending guarantees listed in the FAQ can be broken if you do
sufficiently nasty things with distributed Erlang---IIRC it involved
restarting nodes several times.

I had a go at rewriting that FAQ entry. I got stuck on questions like
"what do you mean by delivery?"  and "did Hans show that the language
doesn't guarantee it, or did Hans just show that the current
implementation can be subverted?" and, finally, "given that you can't
prove (in general) what a process is going to do after delivery,
e.g. it can hang or crash, is guaranteed delivery important?"

I ended up leaving the entry as-is.

Matt