These are chat archives for RBMHTechnology/eventuate

28th
Oct 2016
Dan Di Spaltro
@dispalt
Oct 28 2016 01:21
efficiency wise, do you have to read the whole log/entire event table each time you recover an EventSourcedActor?
Martin Krasser
@krasserm
Oct 28 2016 04:12
No, event logs should index on destinationAggregateIds to make recovery of a single actor efficient (see implementation of Cassandra and LevelDB storage backends). Applications can also save snapshots of actor state or replay from an application-defined sequence number.
Dan Di Spaltro
@dispalt
Oct 28 2016 04:26
@krasserm yeah I think I am starting to understand the nitty gritty details now, if my underlying store handled transactional writes well, I could get away without doing the index writing actor, correct (unlike cass, but similar to leveldb)?
Martin Krasser
@krasserm
Oct 28 2016 04:40
Right, a datastore that does transactional indexing for you can help to simplify the implementation a lot.
Still not similar to LevelDB as it doesn't support secondary indices
Dan Di Spaltro
@dispalt
Oct 28 2016 04:48
right, yeah I see your scheme with the leading bytes
Dan Di Spaltro
@dispalt
Oct 28 2016 07:34
I see it all now, thank you. Ill probably ditch partition since it's "implicit"
@krasserm btw, this is really slick how this all works, feels less magic-y than persistence. Particularly, the eventLog being a straight actor is nice, and the aggregateId is nice.
Martin Krasser
@krasserm
Oct 28 2016 07:38
Thanks, glad to hear that :smiley:
Dan Di Spaltro
@dispalt
Oct 28 2016 07:44
I vote for the cluster example for the next sample project
Martin Krasser
@krasserm
Oct 28 2016 07:46
Note taken
Martin Krasser
@krasserm
Oct 28 2016 12:06
The Akka Streams adapter is finally in master http://rbmhtechnology.github.io/eventuate/adapters/stream.html
Konrad `ktoso` Malawski
@ktoso
Oct 28 2016 12:06
very cool, grats!
Martin Krasser
@krasserm
Oct 28 2016 12:06
Thanks Konrad!
Konrad `ktoso` Malawski
@ktoso
Oct 28 2016 12:07
well used api :) so you don't have to create javadsl/scaladsl :+1:
Martin Krasser
@krasserm
Oct 28 2016 12:07
yep, definitely less effort :smile:
Konrad `ktoso` Malawski
@ktoso
Oct 28 2016 12:08
killer feature which people don't notice very often :]
Martin Krasser
@krasserm
Oct 28 2016 12:09
Add that advise to Akka Streams docs?
Konrad `ktoso` Malawski
@ktoso
Oct 28 2016 12:09
hm yeah we should i guess, for library authors
akka/akka#21750
good hint, thanks
Martin Krasser
@krasserm
Oct 28 2016 12:11
thank you for the fantastic Akka Streams documentation!
and of course the fantastic Akka Streams lib :smiley:
Konrad `ktoso` Malawski
@ktoso
Oct 28 2016 12:12
:D
and same back at you – we're always tracking what new fun stuff you ship in eventuate, very cool things in there :)
Martin Krasser
@krasserm
Oct 28 2016 12:13
the next thing will be an Akka Persistence importer
just kidding :wink:
Konrad `ktoso` Malawski
@ktoso
Oct 28 2016 12:13
lol ;)
half-kidding perhaps though ;)
we'll see what we do about multi datacenter... have to tackle this eventually :)
Martin Krasser
@krasserm
Oct 28 2016 12:22
I'm happy to join the discussions :smiley: Any pointers?
Dan Di Spaltro
@dispalt
Oct 28 2016 19:06
@krasserm could I convince you publish a test artifact?
Martin Grotzke
@magro
Oct 28 2016 20:54
@krasserm awesome, great step!