[erlang-questions] Erlang is the best choice for building commercial application servers

Michael Turner <>
Mon Mar 12 09:51:45 CET 2012


Erlang's main problem might be that simply outlining its virtues
*objectively* only falls on the ears of managers as all-too-typical
geek techno-zealotry. So: perhaps some strategic modesty is in order?
Let me suggest some internal "guerrilla marketing" approaches along
those lines.

- Pitch: "Erlang mainly excels when you use it for a little
orchestration; other languages are usually better for all the
nitty-gritty of choreography, where the rubber really hits the road."
(Yeah, you just mixed at least two metaphors. Guess what? Nobody
noticed.)

OK, team, it's decided: Erlang will just be this relatively thin layer
of relatively substitutable orchestration logic. Very little risk
there, right?

Then, as other parts of the project start to run a bit late ....

- Pitch: "Erlang's kind of slow and piggish for choreography, but
look: with OTP and all, you can throw together *prototypes* of the
choreography very quickly; later, go back and plug in choreography
crafted in the optimal language for it."

OK, team, let's triage: what do we do in Erlang, what do we do in the
originally contemplated languages, and what do we skip (or fake) in
this phase?

Then, when the client loves the demo prototype, and confuses it with a
90% finished product ..."

- Fever Pitch: "Oh crapski, we're running out calendar time on our
schedule, and the customer thinks we're almost done, and we're *much*
more short of resources than we expected to be, at this point in the
project. [Gosh, that *never* happens in software, does it?] That means
we'll have to write a whole lot of distinctly suboptimal code really
fast. It'll have to be concise to a fault, it'll have to draw on all
kinds of libraries, it'll need pretty good tool support ... hmm, I
wonder if ...."

I think this approach works best if the internal champion gives a
strategically-modest presentation on some aspect of Erlang development
about once a week, perhaps in brownbag-lunch style so that managers
don't complain about resources being sucked away during the work day.
At that frequency, he's keeping everybody's brain warmed up for
Erlang, always gaining a little mindshare in the process. Clearly,
this approach also works best if the internal champion is both an
excellent Erlang programmer and very good at coaching others. But how
do you get to that place, except through enthusiasm? And being careful
to *curb* (outward) enthusiasm can be harder than it sounds ....

-michael turner



On Mon, Mar 12, 2012 at 12:05 PM, Richard O'Keefe <> wrote:
> I'm reminded of someone who was writing a program for a US Government
> contract.  The contract required the program to be written in Ada.
> He wanted to write it in Prolog.  Oh dear, what to do?
>
> He wrote a Prolog interpreter in Ada, told them the Ada program was
> the program they had paid for, and this other file with all the :-
> lines in it was an initialisation data file...
>
> I just wish I could remember his name so that I could be sure not to
> mention it (and not get him in retrospective trouble).
>
> Of course, if anyone else had had to maintain the program...
>
> _______________________________________________
> erlang-questions mailing list
> 
> http://erlang.org/mailman/listinfo/erlang-questions



More information about the erlang-questions mailing list