Differences

This shows you the differences between two versions of the page.

Link to this comparison view

irc:1440712800 [2017/05/27 13:44] (current)
Line 1: Line 1:
 +[09:58:41] <purplefox> cescoffier: pmlopes temporalfox morning guys, can you give me 10 mins before we start the meeting?
 +
 +[09:58:49] <temporalfox> ok
 +
 +[09:58:54] <pmlopes> sure
 +
 +[10:00:15] <cescoffier> no problem
 +
 +[12:31:16] *** sendak.freenode.net sets mode: +oo purplefox temporalfox
 +
 +[12:34:41] *** sendak.freenode.net sets mode: +oo purplefox temporalfox
 +
 +[15:32:21] <penthief> Using vertx 3.0.0, vertx.executeBlocking is usually very fast but sometimes it takes a couple of seconds.
 +
 +[19:05:44] <melvinross> if you send a message over the event bus and it's recieved by a vertical on the same instance, is the transport mechanism the same as if it were received by a different instance on the same machine, or within the networked cluster?
 +
 +[19:07:27] <BadApe> i am not technically in a place to answer, but from what i understood, the eventbus once clustered, if you publish all listeners would pick up the message
 +
 +[19:09:05] <melvinross> i'm not asking if they would pick up the message, they do
 +
 +[19:09:46] <melvinross> i'm asking if you did a publish and some consumers were inside of the same JVM instance, and others where attached over the network or inside of a different JVM instance, if the mechanism to send the message is the same
 +
 +[19:09:55] <BadApe> so you what to know if it is a network broadcast?
 +
 +[19:10:18] <BadApe> i assume not until hazelcast has started broadcasting
 +
 +[19:12:29] <melvinross> rather i want to know if hazelcast is involved at all when it comes to delivering event bus messages to verticles running inside the same JVM instance
 +
 +[19:12:56] <melvinross> assuming that vertx has been started in clustered mode
 +
 +[19:13:30] <BadApe> you could double check but looking at your network traffic
 +
 +[19:13:42] <melvinross> that's a good point
 +
 +[19:13:44] <BadApe> however as it is something you have to configure and start up, i would suggest not
 +
 +[19:13:51] <amr> can you use the event bus to send distributed messages?
 +
 +[19:14:02] <amr> as in, via the network
 +
 +[19:14:07] <BadApe> what is a distrubted message? .publish?
 +
 +[19:14:18] <amr> as in, have it arrive on a verticle on another server
 +
 +[19:14:23] <BadApe> sure once you have hazelcast (ha) working, we said that it will pick it up
 +
 +[19:14:37] <amr> oh, i didnt know that!
 +
 +[19:14:40] <BadApe> oh
 +
 +[19:14:42] <melvinross> ha
 +
 +[19:14:44] <BadApe> read up :) we just said it
 +
 +[19:14:53] <amr> oh im bloody blind
 +
 +[19:14:59] <amr> thanks :)
 +
 +[19:15:03] <melvinross> it's not a matter of vertx not working
 +
 +[19:15:10] <BadApe> no problem it is friday, i have beer :)
 +
 +[19:15:57] <melvinross> i'm trying to get my large well known company to adopt vertx as the platform for a new high profile project
 +
 +[19:16:11] <amr> thats actually how i came to learn of vertx
 +
 +[19:16:19] <amr> used internally at a bigcorp
 +
 +[19:16:51] <BadApe> here is my advice, you aren't going to be able to do that, write something small using it and show others
 +
 +[19:17:02] <melvinross> oh i already am :-)
 +
 +[19:17:25] <melvinross> talk is cheap after all
 +
 +[19:17:57] <melvinross> but phase 1 is still effectively a non-demo discussion
 +
 +[19:18:34] <melvinross> from an engineering perspective, vertx is clearly the better choice (compared to falcon)
 +
 +[19:18:58] <BadApe> never heard of it
 +
 +[19:19:03] <melvinross> but there's a lot of "i want to code in python for the sake of coding in python" to fight
 +
 +[19:19:15] <BadApe> you work for bose?
 +
 +[19:19:19] <amr> sounds like jp morgan
 +
 +[19:19:47] <melvinross> http://falconframework.org/
 +
 +[19:19:51] <BadApe> i like python, but imho it is for devops work
 +
 +[19:20:26] <BadApe> not for production platform that have to scale
 +
 +[19:20:33] <BadApe> yeah i am sure there are some good projects out there
 +
 +[19:20:58] <BadApe> ok, question, so i want to run my Verticle from eclipse ide so i can debug etc
 +
 +[19:21:19] <melvinross> are you using maven/grade or not?
 +
 +[19:21:19] <BadApe> i followed the blog about making a fat jar, but that doesn't really help with my dev process
 +
 +[19:21:24] <BadApe> maven
 +
 +[19:21:46] <BadApe> i've not really figured out grade yet, not seen the advantage
 +
 +[19:22:14] <BadApe> ant -> maven, i saw right away, but maven -> grade not seen the advantage to learn it yet
 +
 +[19:23:02] <melvinross> admittedly I actually don't use maven or gradle inside of eclipse and just included the jars since i have vertx installed in my VM
 +
 +[19:24:46] <melvinross> so you can launch from the command line?
 +
 +[19:25:01] <melvinross> if so, then it's just setting up a new run config
 +
 +[20:01:46] <francis_> hello, i've created a verticle, i want to pass in some parameters, when it is deployed i guess i could use new MyVerticle(data)
 +
 +[20:01:55] <francis_> i have a feeling that isn't the right way
 +
 +[20:10:09] <melvinross> what makes you think there's something wrong with using a constructor?
 +
 +[21:15:52] *** ChanServ sets mode: +o purplefox
 +
 +[21:32:14] <BadApe> i have some code that looks like server.connectHandler(socket -> { vertx.deployVerticle(new ClientVerticle(socket)); ...
 +
 +[21:33:29] <BadApe> inside the ClientVerticle there is a socket.handler that tries to deserialise some json, however when there seems to be some extra data being send when a client connects
 +
 +[22:34:29] <amr> actually thats a good point
 +
 +[22:41:57] <amr> guess you cant have the constructor when specifying the string of the class instead of an instance
 +
 +[23:03:53] <amr> oh, pass it in deploymentoptions
 +
 +[23:03:56] <amr> duh
 +
 +[23:33:06] <BadApe> can't see how you would pass in a class constructor value using deployment options