Advice

mike h mike_k_houghton@REDACTED
Wed Dec 16 13:54:52 CET 2009


Thanks for all your comments - I've ordered Joe's book and will no doubt get the OReilly book as well :)

Mike




________________________________
From: Michael Turner <leap@REDACTED>
To: Vasilij Savin <vasilij.savin@REDACTED>; mike h <mike_k_houghton@REDACTED>
Cc: "erlang-questions@REDACTED" <erlang-questions@REDACTED>
Sent: Sun, 13 December, 2009 15:15:29
Subject: Re: Advice


"Armstrong's book is quite heavy in technical details and is more of
reference manual."

I don't know why, but I'm finding it a very pleasant tutorial.  It
reads very smoothly.  It might have more value in some ways as a
reference, but that's clearly not the book's main purpose.

-michael turner


============

Greetings,

We are about to finish our project developing Erlang Cluster and I think
we
actually use something that you are thinking about.

In a nutshell, we have Mnesia for storing data tuples, that is controlled
by
one process and everybody who wants to get a tuple requests it from the
controlling process (no direct access to mnesia). Then this tuple is
assigned in Mnesia transaction and sent to requester.

Mnesia gives you plenty of things for free, so perhaps it is better not to
reinvent the wheel, unless there are good reasons to do so.

Regarding books, can not recommend either. We had I think O'Reillys book
throughout project, but it was not as useful as online documentation for
us.
YMMV. Armstrong's book is quite heavy in technical details and is more of
reference manual.

Hope that helps.

Regards,
Vasilij Savin


On Sun, Dec 13, 2009 at 2:45 PM, Jayson Vantuyl <kagato@REDACTED> wrote:

> Honestly, you might do better to just use Mnesia as the backend for a
> tuple-space. It's distributed and has all of the features.
>
> Otherwise, I wouldn't probably make each tuple is a process. If
> distribution isn't a goal, you might consider using ets tables. They're
> lightning fast and you can do pretty powerful queries on them using
> ets:match and friends.
>
> I can't currently think of a big advantage to using processes as tuples,
> although I admit that the objects-as-processes model is a good way to move
> from OOP to FP.
>
> Good luck.
>
> On Dec 13, 2009, at 4:09 AM, mike h wrote:
>
> > Hi All,
> >
> > I've just started learning Erland and I'm really excited by it!
> > I have a couple of questions.
> >
> > I have a lot of experience in OO languages and quite a bit in some of the
> imperative/functional hybrids like Ruby etc. My main interest in Erlang is
> in it's concurrent and distributed capabilities, so, which of the two main
> Erlang books would folk reccommend? (I think I'm right in saying there are
> currently just two main books, one by Oreilly and one by Erlang's designer)
> >
> > Next question is a bit more open ended. I have a particular interest in
> tuple spaces (JavaSpaces, Linda, Rinda etc) and their implementation and
> application.. And I want to use Erlang to implement a tuple space server and
> client. At the moment I've not really yet got into the Erlang 'mindset' but
> my initial idea is to use a process to represent a Tuple and have one
> process as the space server and a message to that server would create a
> tuple (ie process) based on the message details.
> >
> > Any comments, advice, pointers to existing implementations of a tuple
> space etc etc would be welcome,
> >
> > Thanks
> >
> > Mike
>
> --
> Jayson Vantuyl
> kagato@REDACTED
>
>
>
>
>
>



      


More information about the erlang-questions mailing list