>
> A related question. If I wanted to do a once-off initialization of the value
> (ie, the first VM to create the distributed ref will set the value) how would
> I go about it?
>
I would just create the Ref without an initial value, and if the current value
is non-nil, you know that it has
On 3 December 2011 01:44, David Edgar Liebke wrote:
> Hi Glen,
>
> >
> > The init-stm step is still referenced in the documentation as being
> required BTW.
> >
>
> Thanks, I'll remove the reference.
>
> > I had a couple of questions.
> >
> > I noticed that when I create a reference (zk-ref) I ne
Hi Glen,
>
> The init-stm step is still referenced in the documentation as being required
> BTW.
>
Thanks, I'll remove the reference.
> I had a couple of questions.
>
> I noticed that when I create a reference (zk-ref) I need to provide an
> initial value. For each VM I do this for - it en
On Fri, Dec 2, 2011 at 12:33 AM, Glen Stampoultzis wrote:
> I noticed that when I create a reference (zk-ref) I need to provide an
> initial value. For each VM I do this for - it ends up clobbering the
> previous value. Is there anyway to create a reference without necessarily
> clobbering exis
On 2 December 2011 05:26, liebke wrote:
> Just released Avout 0.5.2, which now includes automatic STM
> initialization (no more pesky init-stm step).
>
>
The init-stm step is still referenced in the documentation as being
required BTW.
I had a couple of questions.
I noticed that when I create a
On 1 Dec 2011, at 18:26, liebke wrote:
> Just released Avout 0.5.2, which now includes automatic STM
> initialization (no more pesky init-stm step).
Ha, throw down the gauntlet, then beat me to it ;-)
Great work,
Sam
---
http://sam.aaron.name
--
You received this message because you are sub
Just released Avout 0.5.2, which now includes automatic STM
initialization (no more pesky init-stm step).
David
On Dec 1, 12:44 pm, David Edgar Liebke wrote:
> init-stm (and reset-stm) are only used for creating an STM that will be used
> by every client. You only want to call it once, not ever
init-stm (and reset-stm) are only used for creating an STM that will be used by
every client. You only want to call it once, not every time a client connects.
I could be more clever about calling it when it's clear that it hasn't been
called before though. I'll dedicate that patch to you :)
> I
Out of interest, why is #'init-stm a separate step to #'connect
I tried looking at the docstrings for each fn but they were both nil :-(
Sam
---
http://sam.aaron.name
On 1 Dec 2011, at 17:21, David Edgar Liebke wrote:
> Did you initialize the STM?
>
> (init-stm client)
>
> You only need to d
Fantastic, I've added the init-stm step to the code snippet at the top of the
Avout site, it was an accident to leave it out.
It's been great "pairing" with you two :)
David
> Works perfectly for me too. Perhaps it might help to add that to the example
> snippet to stop idiots like myself fall
On 1 Dec 2011, at 17:21, David Edgar Liebke wrote:
> Did you initialize the STM?
>
> (init-stm client)
Works perfectly for me too. Perhaps it might help to add that to the example
snippet to stop idiots like myself falling into that trap :-)
Sam
---
http://sam.aaron.name
--
You received th
That's the ticket! Thanks David, its working for me now.
On 01/12/2011 17:21, David Edgar Liebke wrote:
> (init-stm client)
--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts
Did you initialize the STM?
(init-stm client)
You only need to do it the first time, to set up the necessary zookeeper nodes,
it's described in the main tutorial but not the snippet on the top of the avout
site.
David
On Dec 1, 2011, at 12:15 PM, Sam Aaron wrote:
> Hi David,
>
> thanks for
Hi David,
thanks for looking into this so promptly. Sadly 0.5.1 just throws a different
exception:
user=> (def client (connect "127.0.0.1"))
#'user/c
Thanks Sam and Edmund,
The stack traces were helpful, I think I understand what the immediate problem
is. It appears that the transaction ID in these cases is not getting set, and
then Avout is trying to write data to the ZooKeeper node /stm/history/ instead
of /stm/history/txid.
Since I can'
Hi David,
I nuked all my zookeeper deps in my lib and ~/.m2 dirs, but similar to Edmund
experience it doesn't fix anything. My stacktrace is also identical:
∴ /Users/sam/tmp/avv
λ
Hey Sam & Dave,
I'm following along at home. I followed David's advice and there's
no change. FWIW here's the client datastructure and the exception
stacktrace:
#
java.lang.IllegalArgumentException: Path must not end with / character
at org.apache.zookeeper.common.PathUtils.validatePat
Hey David,
I get an identical exception here here w/ zookeeper version
3.2.2. Because I'm crazy that way I also tried to call the ref "/r0/"
(just to see) and the exception came up. However it was different in
that with "/r0/" I got the usual exception handling in emacs, whereas
with "/
Hi Sam,
> run-in-transaction exception: # java.lang.IllegalArgumentException: Path must not end with / character> nil
Very interesting, I wouldn't expect that particular exception unless you named
the zk-ref "/r0/" instead of "/r0", which you apparently didn't.
And even when I do call it "/r
Hi David,
I'm super excited by Avout. It seems *better* than magic in that it not only
appears to make complicated things possible, but also in a conceptually
transparent way. Crazy cool.
I'm about to look into this in detail, but I thought I'd just post an issue I'm
having with the basic exam
On Wed, Nov 30, 2011 at 2:39 AM, Linus Ericsson
wrote:
> In Clojure in Action (still in MEAP i think) there's a chapter about
CiA was officially released at the Conj. Everyone who bought the MEAP
should have had a notification by now about downloading the final
version (my final CiA eBook is date
In Clojure in Action (still in MEAP i think) there's a chapter about
using (Erlang/OTP-based) RabbitMQ message queue server for making
Clojure scalable in a Hadoopish map-reduce-way.
Avout seems to solve many of the problems that easily could occur in
such an approach by not needing a centralized
If you're willing to dig into another language, 'Erlang and OTP in
Action' gives a great overview of Erlang and the distributed
principles underlying that language. Though different from the
approach of distributed STM, the concepts of distributed applications
are baked into the core of Erlang.
On
Could anyone please recommend a good introductory book about
distributed application development? The release of Avout has gotten
me interested in the subject.
Thanks,
Harrison.
On Nov 29, 7:38 pm, liebke wrote:
> Today we are releasing Avout, which brings Clojure's in-memory model
> of state to
Fantastic !!! I'm looking forward to giving this a try. It has
the potential to solve some problems I am currently working on.
so thanks for your efforts.
Tim
On Nov 29, 10:38 am, liebke wrote:
> Today we are releasing Avout, which brings Clojure's in-memory model
> of state to distributed
Congrats on the release! Looks like the world just got a bit more civilized :)
Particularly excited to see how far this concept of distributed refs
can go while remaining simple:
- Using S3 as the backing store
- Massively distributed STM. For example, every user of clojure
sharing datastructure
1) On the avout.io site, is the diagram of conflicting transactions
> correct? It looks to me like the red arrow is in the wrong place (and it
> doesn't match the description below it, points 5 and 6).
>
Great catch, I had intended to fix the figure before release but forgot.
It's fixed now.
The issue with transactions not overlapping with in-memory ones implies
some separation to deal with distributed coordination, I think. Are there
any guidelines or interesting papers on how to create an effective
distributed architecture with these semantics?
--
You received this message beca
Wow. It will take a while to digest this before I can even dream of what
possibilities this opens up.
In the meantime, a couple of simple questions:
1) On the avout.io site, is the diagram of conflicting transactions
correct? It looks to me like the red arrow is in the wrong place (and it
do
Today we are releasing Avout, which brings Clojure's in-memory model
of state to distributed application development by providing a
distributed implementation of Clojure's Multiversion Concurrency
Control (MVCC) STM along with distributable, durable, and extendable
versions of Clojure's Atom and Re
30 matches
Mail list logo
