The chat today was about deployment descriptors and some distribution issues regarding the upcoming M1 release.
Note - I missed 7 minutes (from 19.24 to 19.31) due to connection problems [18:54] *** Axis2 Developer Discussion [18:54] *** [freenode-info] why register and identify? your IRC nick is how people know you. http://freenode.net/faq.shtml#nicksetup : #apache-axis [18:55] *** Deepal joined #apache-axis [18:55]<Harsha> Glen: Did u enjoy your recent visit to Melbourne? [18:56]<gdaniels> I did! Sorry we didn't get a chance to hook up while I was there... [18:56]<gdaniels> Hopefully I'll be down there again at some point [18:56]<Harsha> That's Ok. Maybe the next time!! [18:56]<gdaniels> :) [18:58] *** Chamil joined #apache-axis [19:00]<Deepal> hi Harsha [19:01]<Ajith> Hello all [19:01]<Deepal> thnaks againg for your comments on Arch guide and client API [19:01]<Srinath> Hi All; Glen u r early :) [19:02]<Harsha> Hi All [19:02]<gdaniels> Hi Srinath - woke up early this morning :) [19:02]<Srinath> :) [19:02]<gdaniels> Harsha : did I miss those comments? When did you send them? [19:03]<Deepal> Glen nope [19:03]<Harsha> Glen: I sent them directly to Deepal and Ajith. [19:03]<Deepal> he sent directly to me [19:03]<Ajith> Oh Harsha mailed personaly [19:03]<Ajith> BTW I updated the docs in SVN [19:03]<Harsha> I am sorry. I don't know the protocol for reviews. [19:04]<Deepal> me too :) [19:04]<Ajith> And commented that it was reviewd by Harsha [19:04]<gdaniels> oh, ok - although in general good to have that stuff archived, IMHO [19:04]<Srinath> and there is a site up too I think at chintka's site [19:04]<Ajith> You mean chinthaka's account [19:04]<Ajith> :) [19:06]<Ajith> http://www.apache.org/~chinthaka/axis2/ [19:06]<Ajith> this is the maven generated site [19:06]<Ajith> as of today [19:07]<Srinath> where are we in the term of M1? [19:07]<Srinath> I think funcitionlitywise we are quitre there IMHO [19:07]<Ajith> Well I guess the docs need to be reviewd [19:07]<Deepal> yep Ajith [19:08]<Deepal> Glen : how about the code review :) [19:08]<gdaniels> I haven't looked at it much more since last week, unfortunately - too much other stuff on my plate :( [19:09]<gdaniels> But I don't want to hold things up either [19:09]<Deepal> cool [19:09]<Srinath> glen: got a Q for you, how you write the xsd for the wsdd to handle the <handler ref=".." > and <handler>...</handler> [19:09]<gdaniels> I wish a few others were looking closely at it too [19:09]<Srinath> in Axis 1.x [19:09]<gdaniels> I mean I know all you guys are, but you pretty much share a brain anyway :) [19:10]<Srinath> :) [19:10]<Ajith> ofcourse [19:10]<Ajith> More brains the better it is :) [19:10]<Srinath> brain are never too much [19:10]<gdaniels> :) [19:10]<Deepal> :D [19:10]<gdaniels> It's <handler name="foo" type="java:classname"/> to define, and <handler type="foo"/> to ref [19:11]<Deepal> glen : u mean Axis 1.x or axis2 [19:11]<gdaniels> 1.X [19:11]<Deepal> oki [19:11]<Srinath> how you sepcify it in the xsd [19:11]<Srinath> ? [19:11]<gdaniels> just by defining the two attributes [19:11]<gdaniels> xsd doesn't get you all the co-occurrence constraints you might want [19:12]<Srinath> yah [19:12]<gdaniels> RelaxNG does a lot better there [19:12]<Srinath> with all this complexity it is not expressive enough :( [19:12]<gdaniels> heh [19:12]<Srinath> what to do facts of life [19:13]<gdaniels> Well we know we're going to have application-level validation anyway to check references and such, so I'm generally not worried about adding those kinds of checks too [19:13]<Srinath> may be (after M1 we should go for xmlbens for DD but let pout it to future) [19:13]<gdaniels> I think <handler name=""> to define and <handler ref=""> to ref is good [19:14]<gdaniels> the problem is that you need to determine scoping [19:14]<gdaniels> which is why we didn't end up with that in Axis1 [19:14]<Srinath> ok glen let live with a another attributes ref [19:15]<gdaniels> So when you ref a handler, do you mean a single instance, or just the type? [19:15]<Srinath> they are stateless so do not matter? [19:16]<gdaniels> Oh right we're doing 100% stateless now [19:16]<gdaniels> ok [19:16]<Srinath> yap [19:17]<Srinath> glen:should we define transport recivers and transport senders in the dd? [19:17]<Srinath> (just curious 4 the futrue) [19:17]<Deepal> we have alredy done :) [19:17]<gdaniels> Why wouldn't you? [19:18]<Deepal> am I correct Srinath [19:18]<Srinath> they are not define in dd <transport> now [19:18]<Srinath> it is like <transport name=""> <parameter/></transport> [19:19]<Srinath> may be we should put <transport name="" reciver=".." sender="..."/> [19:19]<gdaniels> s/reciver/receiver/ [19:19]<Srinath> yap :D oops [19:19]<gdaniels> or <sender name=""> <receiver name=""> [19:19]<Srinath> sure [19:20]<gdaniels> <sender name="" protocol=""> ? [19:21]<Srinath> I do not get it glen [19:21]<gdaniels> <sender name="MySpecialWebThing" protocol="http"> [19:21]<Srinath> at what level inside tranport or out? :) [19:21]<gdaniels> to separate the namespace of senders from the protocols [19:22]<gdaniels> but I don't know if we want to do that [19:22]<gdaniels> Srinath: that would define a "client transport" [19:22]<gdaniels> In Axis1, we use the name of the transport as the protocol in the URL [19:24]<Ajith> hmmmm Would not the transport be specified in the WSDL? [19:24]<Ajith> I mean the endpoint [19:31] *** Axis2 Developer Discussion [19:32]<Ajith0213> Ah finally [19:32]<gdaniels> Just like if you're coming out of the authentication phase, let's say, and you don't have a security principal set, that would be an error [19:32]<Srinath_> yap .. I remeber we say most of them are static tests but here may be daynamic tests as well [19:32]<gdaniels> wb Ajith :) [19:32]<Srinath_> I got it clean glen: how to implement it [19:32]<gdaniels> Srinath: right [19:32]<gdaniels> look at EngineToy :) [19:32]<Srinath_> are we let ppl extends the phases? [19:32]<gdaniels> yes [19:32]<Srinath_> will check cool:) [19:33]<gdaniels> when you define a phase, you can optionally tell it what class to use [19:33]<gdaniels> if you don't tell it, it's a "raw" Phase with no checks. Otherwise it's your own class. [19:33]<gdaniels> That can also allow for other extensions in the future [19:33]<Srinath_> cool +1 [19:34]<gdaniels> So why can't I get a simple OM serialization example working? [19:34]<gdaniels> Prints nothing on my screen :( [19:34]<Srinath_> flush it :D [19:34]<Ajith0213> ooops [19:34]<Srinath_> the stream we all got in tthat trouble [19:35]<Ajith0213> I did not update the code in the OM tute to have the flush statement [19:35]<gdaniels> i.e. System.out.flush() at the end? [19:35]<Ajith0213> Sorry Glen : I will correct the code [19:35]<Ajith0213> Nope [19:35]<Ajith0213> Writer.flush() [19:35]<Srinath_> you got to flush System.out .. (bit werid but ..) [19:35]<gdaniels> ah [19:35]<gdaniels> got it [19:36]<gdaniels> much better :) [19:36]<gdaniels> thanks Ajith, that would be great to fix :) [19:37]<Srinath_> should we change the transports for M1 [19:37]<Srinath_> ? [19:37]<gdaniels> Could we think about having a pretty/nonPretty switch somewhere for serializing/debugging OM? [19:38]<Ajith0213> oh [19:38]<Ajith0213> We did not include anything as such [19:38]<gdaniels> Maybe a different serializer, but I'd want to be able to select it without changing code [19:38]<gdaniels> I'm saying for the to-do list, not for today :) [19:39]<Ajith0213> But I guess we can have a slightly modified writer to do that [19:40]<Srinath_> may be create it via a factory and boolen flag to switch [19:40]<Ajith0213> We are currently using the streamwriter in the RI [19:40]<gdaniels> right - although it would be nice if there were a scoped config type setting that I could just switch on via an environment variable or something [19:41]<gdaniels> Oh, so this is a stax issue then [19:41]<Ajith0213> We can actually write one of our own to do that [19:41] *** Srinath quit FreeNode : Read error: 113 (No route to host) [19:41] *** Chamil quit FreeNode : Read error: 113 (No route to host) [19:41]<gdaniels> we're just pushing API calls to them, not actually writing the string ourselves [19:41]<Ajith0213> Yep [19:42] *** Ajith quit FreeNode : Read error: 113 (No route to host) [19:42] *** Deepal quit FreeNode : Read error: 113 (No route to host) [19:42]<gdaniels> hmm [19:42]<Srinath_> yap glen [19:42]<gdaniels> ok, this warrants a little research then [19:42]<Srinath_> (it is our old selves quiting .. do not worry :D ) [19:43] *** Ajith0213 is now known as Ajith [19:43] *** Deepal0015 is now known as Deepal [19:44]<Srinath_> what else we have todo's for M1 ? [19:44]<Srinath_> when will we move the code to src tree? [19:45]<Ajith> Hmmm BTW what about the docs [19:45]<Ajith> ? [19:45]<gdaniels> I thought at one point we discussed changing the dir structure, btw... [19:45]<gdaniels> java/src, not src/java [19:45]<Srinath_> yap glen [19:46]<Ajith> I mean are they technically sufficient [19:46]<gdaniels> so you wouldn't end up with src/schema, src/samples [19:46]<Ajith> ? [19:46]<gdaniels> but you'd have java/samples, etc [19:46]<Srinath_> mm .. glen in maven sense src/schema's is fine [19:46]<gdaniels> Ajith: Haven't read them in enough detail yet, so I can't say. OM one looked good tho [19:47]<Ajith> oh cool [19:47]<Srinath_> but I am not pressing it .. lets take one appoch and stick to it [19:47]<Ajith> we still have to fix a bit in the archi guide and the user guide [19:47]<Srinath_> glen:I accept u r concerns having java/src/java [19:48]<Ajith> and somebody suggested a small getting started guide for the developers [19:48]<gdaniels> Srinath_: ok - are you good with java/ at the top level (in parallel with c++/ later...) and then src/, samples/, test/, doc/, etc under that? [19:49]<Ajith> Hmmmm [19:49]<Srinath_> fine [19:49]<Ajith> Cant we not share samples and all [19:49]<gdaniels> ? [19:49]<Ajith> at least some of the documents [19:49]<Ajith> then we need some higher directory to keep the shared items [19:50]<gdaniels> oh [19:50]<gdaniels> so you mean axis2/docs, axis2/schemas, and then axis2/java/src, axis2/java/samples, etc [19:50]<Ajith> exactly [19:51]<gdaniels> we'd still want a separate java/docs for instance, but for truly shared stuff I'd be fine with that [19:51]<Ajith> yep [19:51]<Srinath_> yes otherwise we might get in to trouble politically :D [19:51]<Ajith> :D [19:51]<gdaniels> I don't think there's going to be that much that's deeply shared, though, honestly - just a few docs and schemas [19:52]<gdaniels> but +1 for setting it up [19:52]<Srinath_> cool shall I move it to src tommarow then? [19:52]<Ajith> seems ok to me [19:52]<gdaniels> +1 [19:53]<Ajith> BTW what about the distribution [19:53]<Ajith> I mean what do we include there [19:54]<gdaniels> The distribution should look like Axis2/docs, Axis2/webapps, Axis2/samples, Axis2/lib, etc [19:54]<Ajith> docs,webapps,samples and lib ? [19:54]<Srinath_> axis2 or axis [19:54]<Ajith> BTW do we need a lib ? [19:55]<Deepal> lets use as axis2 [19:55]<Ajith> I mean we can just have a war [19:55]<Srinath_> since maven we do not need a lib I think [19:55]<gdaniels> I think it should be axis2, at least for now [19:55]<Ajith> ah yes [19:55]<gdaniels> and why not a lib? [19:55]<Ajith> thats also a point [19:55]<gdaniels> where do the jars go? [19:55]<Srinath_> we do not need to checked in any ljars [19:56]<Srinath_> as they are downlaoded by maven' [19:56]<gdaniels> we're talking about the distribution here [19:56]<Srinath_> sorry oops [19:56]<Srinath_> :( [19:56]<gdaniels> we're not going to force people to use maven in order to use axis?! [19:56]<gdaniels> ok :) [19:56]<Ajith> the thing is glen if you look at the current dist, web-inf/lib and lib are identical [19:57]<Srinath_> about dist :) are we givin g a war inside the dist [19:57]<gdaniels> Ajith: doesn't matter. Axis2 should be embeddable and I should be able to use it even without a servlet engine [19:57]<Ajith> Ok agreed [19:57]<gdaniels> I rarely use Axis inside a servlet engine myself - I use SimpleAxisServer all the time [19:57]<Ajith> ! [19:58]<gdaniels> I like embedding it into other programs without the weight of Tomcat/Jetty, and it's easier to test too [19:58]<Srinath_> shall we not give a war ? [19:58]<gdaniels> No, we should definitely have a war [19:59]<gdaniels> I'm just saying we should also have a lib/ is all. :) [19:59]<Ajith> sure [19:59]<Srinath_> yap it make the dist large by repeat the jars [19:59]<gdaniels> If we want to make the distribution smaller, I think adding a build.xml which builds the war from the lib/ is OK [20:00]<gdaniels> most java people have ant these days - not yet maven tho [20:00]<Srinath_> (I remeber at unv I hate big dist as it take ages to download) [20:00]<Srinath_> +1 for build.xml [20:00]<gdaniels> sure, +1 [20:00]<Srinath_> what should be put in the docs? [20:01]<gdaniels> a bunch of english text which tells people how to do stuff. [20:01]<gdaniels> :) [20:01]<Ajith> I dont know , I would rather like to see a ready to use jar in the dist [20:01]<Ajith> I mean war [20:02]<Srinath_> :) ahould we give java docs .. I think nyes [20:02]<gdaniels> Let's take a VOTE on the list for this one then, to get all the folks who aren't here [20:03]<gdaniels> either 1) include setup for standalone and build war with ant, or 2) include war and build standalone dirs with ant [20:03]<gdaniels> I gotta run, folks - meeting coming up [20:04]<Ajith> :) [20:04]<Ajith> Ok [20:04]<Srinath_> :) quite niceway out [20:04]<Ajith> See ya [20:04]<gdaniels> Well, not really... :/ [20:04]<Srinath_> bye glen and thanks [20:04]<Ajith> :) [20:04]<gdaniels> later all, talk to you soon [20:04]<Ajith> Hey he was just joking [20:04] *** gdaniels quit FreeNode : [20:05]<Srinath_> opps nice way out menns about the dist solution [20:05]<Srinath_> not glen runnig away [20:06]<Ajith> Ah I thought you were joking about glens hurry to leave [20:06]<Srinath_> :) [20:06]<Ajith> BTW lets finish it for the day guys [20:06]<Srinath_> yap I think [20:06]<Ajith> I will send the log (even though I missed a bit in the middle) [20:06]<Harsha> Ok. Thanks Guys. Until Later. [20:07] *** Harsha quit FreeNode : -- Ajith Ranabahu
