Differences

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

Link to this comparison view

— irc:1431640800 [2017/05/27 13:44] (current)
Line 1: Line 1:
 +
 +
 +[00:12:42] <jtruelove> has anyone noticed that the failureHandler on the 'all' route doesn't seem to work https://gist.github.com/jtruelove/a12173eb5dd9fc10960c
 +
 +[00:13:18] <jtruelove> the loggerHandler logs the missed resource BUT the failure handler doesn't
 +
 +[16:35:12] <purplefox> temporalfox: hi julien, how's it going?
 +
 +[16:35:21] <temporalfox> purplefox hey almost the weekend :-)
 +
 +[16:35:42] <purplefox> yay! Although I suspect I will be working most of it :(
 +
 +[16:35:43] <temporalfox> purplefox doing good, wrapping up work
 +
 +[16:36:16] <temporalfox> purplefox I will probably work too, although I do have duties that will keep me busy
 +
 +[16:37:24] <temporalfox> I have almost done with ruby / metrics / http service factory
 +
 +[16:37:47] <temporalfox> after that I'll wrap the List<@DataObject> PR and implement the missing languages
 +
 +[16:38:24] <temporalfox> after that, do you want me to handle something in particular ?
 +
 +[16:38:54] <purplefox> i guess anything with your name against it on the roadmap
 +
 +[16:39:53] <temporalfox> ok
 +
 +[16:40:01] <purplefox> pmlopes: thanks for looking at the auth stuff. looks like we posted our msgs at the exact same minute ;)
 +
 +[16:41:05] <pmlopes> purplefox, i just updated my fork to use your base
 +
 +[16:42:20] <pmlopes> but there are some compilation issues, also what is the purpose of dehydrate/rehydrate i guess it is a fast serialization, but for jwt, it is a plain jsonObject, is there a helper for that or do i need to iterate the keys myself?
 +
 +[16:44:40] <purplefox> that's basically for the case where the User objects needs to be serialized so it can go in the clustered session and be available on other nodes
 +
 +[16:45:25] <purplefox> now for jwt not sure you would need to implement that as the jwt token would be passed on each request i guess
 +
 +[16:46:14] <pmlopes> yes so in that case i should say no clustable and my jwt user does not implement the clusterSerializable interface
 +
 +[16:47:00] <purplefox> yeah more or less
 +
 +[16:47:19] <purplefox> i'm still figuring out the serialization stuff, it's a bit awkward
 +
 +[16:50:14] <purplefox> i'm thinking maybe there should be a subclass of AuthProvider called ClusterableAuthProvider so we don't pollute AuthProvider with the methods which are only used for the clusterable ones...
 +
 +[16:51:03] <pmlopes> ok, i will disable it for JWT, now still need to work out the generation of tokens, do you agree on a method on AuthProvider, like: void getToken(JsonObject options, Handler<AsyncResult<JsonObject>> resultHandler);
 +
 +[16:52:25] <pmlopes> using this api we can generate tokens for jwt, oauth and even refresh tokens since the refresh would be a property of the options json
 +
 +[16:53:08] <purplefox> I suppose this would be on a subclass of AuthProvider not AuthProvider itself as not all AuthProviders will use tokens
 +
 +[16:54:06] <pmlopes> ok, i will implement it
 +
 +[17:02:19] <temporalfox> purplefox can you validate my metrics PR ?
 +
 +[17:04:47] <purplefox> temporalfox: question... why does metrics.close have to be run on a context?
 +
 +[17:05:22] <temporalfox> so the callback to the metrics SPI is done with the same thread/context than the one it was created before
 +
 +[17:06:39] <purplefox> ok, but why is that important?
 +
 +[17:07:11] <temporalfox> for consistency
 +
 +[17:36:12] <jtruelove> purplefox: here's a gist on the failure handler issue https://gist.github.com/jtruelove/7a1a4021b1acf0960aa4
 +
 +[17:39:48] <purplefox> jtruelove: what are you expecting to happen in this example, I'm not sure I follow...
 +
 +[17:40:59] <jtruelove> i'm expecting the failure handler to be called
 +
 +[17:41:30] <jtruelove> because there's no match setup for that
 +
 +[17:42:00] <purplefox> and what actually happens?
 +
 +[17:42:10] <jtruelove> nothing
 +
 +[17:42:27] <jtruelove> the logger handler does get called but not my failure handler
 +
 +[17:42:55] <purplefox> and what happens if you remove your failure handler?
 +
 +[17:43:28] <purplefox> you should get a 404 anyway
 +
 +[17:43:41] <purplefox> so i'm not sure your failure handler is adding anything here
 +
 +[17:43:41] <jtruelove> oh yeah you do
 +
 +[17:44:02] <jtruelove> well the point is to show that it isn't working not to show the value :)
 +
 +[17:44:34] <jtruelove> normally i might track metrics here around failed requests for unknown resources etc..
 +
 +[17:45:03] <jtruelove> if you want to know when people are asking for things that don't exist this is a way to do thaht
 +
 +[17:45:21] <purplefox> i think what is happening is... the failure handler will only be called if the request isn't handled by a normal handler, but in this case it is being handled by the logger handler
 +
 +[17:47:17] <jtruelove> no, that isn't the case. I take at the handler line and the same thing happens.
 +
 +[17:47:31] <jtruelove> at = out
 +
 +[17:48:15] <jtruelove> it returns a 404 by default of course but does not hit my failure handler
 +
 +[17:52:21] <purplefox> you should be able to get what you want by adding a normal handler as the last handler, but if you could add an issue for this, someone can take a look
 +
 +[18:27:53] <jtruelove> sounds good, thanks!
 +
 +[18:36:08] <jtruelove> here's the bug https://bugs.eclipse.org/bugs/show_bug.cgi?id=467420
 +
 +[18:36:17] <AlexLehm> i have the strangest issue when running the unit tests in a CI system. I have added a warning when a test takes more than 2 seconds and the longest a test takes is 10 seconds since it waits for a timeout. when I run the same tests on CI some tests take 80 seconds
 +
 +[18:36:31] <AlexLehm> and the happens on jenkins as well as on travis-ci
 +
 +[19:44:14] <jtruelove> i tested ssl with websockets yesterday with vertx3, setting up your own cert and making the browser accept it is a real pain
 +
 +[23:04:48] <spriet2000> good evening
 +
 +[23:07:18] <AlexLehm> what are you using to write asciidoc files? I currently use atom.io, which has quite a good preview but that doesn't support link: for example
 +
 +[23:08:40] <spriet2000> sublime currently
 +
 +[23:08:57] <spriet2000> i will give atom a try looks nice
 +
 +[23:12:21] <spriet2000> Alex how is your smtp module?
 +
 +[23:12:33] <AlexLehm> the preview function is actually quite good
 +
 +[23:13:01] <AlexLehm> i am working througg the suggestions Tim gave, actually its coming along quite well
 +
 +[23:15:09] <spriet2000> whats the difference between https://github.com/vert-x3 / https://github.com/vert-x3/vertx-mail-client and yours?
 +
 +[23:15:30] <spriet2000> ah
 +
 +[23:15:43] <spriet2000> its moved cool
 +
 +[23:16:11] <AlexLehm> mine is just a fork now
 +
 +[23:16:28] <AlexLehm> mostly to a CI job before pushing it to the read repository
 +
 +[23:18:11] <AlexLehm> irc really needs an edit function
 +
 +[23:18:23] <AlexLehm> mostly to a do CI job before pushing it to the real repository
 +
 +[23:18:25] <spriet2000> :>
 +
 +[23:21:18] <spriet2000> how do you test to find out if code you wrote is fast?
 +
 +[23:22:22] <AlexLehm> i have not done any performance testing yet, its not triggering the blocking checks, so it should be ok
 +
 +[23:24:07] <spriet2000> i want to get some skills for performance testing but its difficult
 +
 +[23:24:40] <spriet2000> i am not jvm expert .. is jmh a good tool to use?
 +
 +[23:25:29] <spriet2000> i will plan a session for it to see if its getting informative feedback
 +
 +[23:27:53] <AlexLehm> actually I think I have not done any projects who really had to do benchmarks
 +
 +[23:28:16] <spriet2000> me neither.. but when is code good?
 +
 +[23:28:44] <spriet2000> i really need some references