Session Start: Tue Jun 23 07:45:09 2009 Session Ident: #apache-wink 03[07:45] * Now talking in #apache-wink 01[07:45] morning all [07:45] hi 03[07:45] * nick is now known as Guest76591 [07:45] guess someone has the name nick 01[07:45] haha... 01[07:46] isn't there a command I can use to change my name? [07:46] use /nick [07:46] "/nick" i think 03[07:46] * Guest76591 is now known as ngallardo [07:46] /nick 01[07:46] thanks Michael 03[07:47] * elib (n=elib@minotaur.apache.org) has joined #apache-wink [07:49] nick: btw, it forced you to change the nick, because nick "nick" is registered to someone 03[07:53] * nfischer (i=c0762d02@gateway/web/freenode/x-761722173e3d79d0) has joined #apache-wink [07:54] Hi everyone [07:55] Hi Nadav 01[07:55] Hi Nadav 01[07:55] ah... thanks Michael. I'll just change it in my config so it sticks. 01[07:55] I think we should have a few more folks joining still 03[07:56] * rott (n=rott@32.97.110.56) has joined #apache-wink 03[07:57] * elib_ (i=c0762d02@gateway/web/freenode/x-ba414dc9475da041) has joined #apache-wink 02[07:58] * martinsnt (n=martinsn@minotaur.apache.org) Quit (Remote closed the connection) 02[07:58] * elib (n=elib@minotaur.apache.org) Quit (Remote closed the connection) [07:59] \help 01[07:59] is the server just dropping you? 03[07:59] * elib_ is now known as elib [08:01] we tried tunneling, since IRC is blocked by the corporate firewall [08:01] but it drops the connection after a while [08:01] so we moved to web client 01[08:02] ah, cool 01[08:02] let's wait for martin to jump back on and we can get started... 03[08:03] * jdillon (n=jdillon@58.147.49.227) has joined #apache-wink [08:03] hi all 01[08:04] hi Dims 01[08:04] hi Jason [08:04] hi dims 01[08:04] is Martin joining again? [08:05] yep 03[08:05] * martinsn (i=c0762d02@gateway/web/freenode/x-6182aace87ed39e6) has joined #apache-wink [08:05] hi all 03[08:06] * gtruty (n=gregtrut@32.97.110.64) has joined #apache-wink 01[08:06] welcome back Martin. alright... let's get started. 01[08:06] thanks to everyone that added content to the agenda 01[08:06] http://cwiki.apache.org/confluence/display/WINK/Architecture+Meeting+And+Agenda 01[08:07] I know that one of the main goals of this discussion is to get the issues related to the two runtimes out on the table in order to pick a base. 01[08:07] If it's all the same to everyone else (mostly the HP team), I'd like to see if we can save ourselves a little time there. 01[08:08] We done some analysis on our end and are very comfortable using the HP code as a base for the wink project. 01[08:08] Now, granted... We cannot formally make that decision in this chat. It must be put up for a vote by the community. 01[08:09] But, in an effort to make the best use of our time here, can we make the assumption that we'll go with the HP runtime and then have a discussion about where we want to get started from there? 01[08:09] Again, sorry to usurp the previous agenda. I think some of it still holds, and we can go through the runtimes still. 01[08:09] Just wanted to throw that out there though. 01[08:09] Thoughts? Comments? 01[08:10] [08:11] Sounds good. If anyone has questions regarding HP runtime, we may answer it here 01[08:11] Thanks Michael... there are a few things we would like to discuss. [08:11] I still have not looked at the HP runtime yet, but if you, nick, are comfortable going with it for the main wink impl seems I have no objections 01[08:11] Cool, thanks Jason [08:11] can we adapt any of the testsuite from the jaxrs to suit the hp runtime at all? 01[08:12] From our perspective, the HP code is more mature and in a better state for modularization. 01[08:12] yes, absolutel.y 01[08:12] and I think that's one of the areas where we can draw the most value from the IBM code. [08:12] Do we have any idea how performant the hp codebase is? [08:12] I was planning on doing much of that work -- the porting of IBM testcases to HP base 01[08:12] in addition to the TCK compliance, we've added tests for a number of scenarios that are not captured by the certification. [08:13] i like the idea of adapting all existing tests from ibm impl and getting them to work on HP base [08:13] we will need to do some code changes to make JAX-RS compliant (subresource locators, minor things mentioned in the JavaDoc but no where else, etc. ) but nothing that's alarming 03[08:13] * dkulp (n=dkulp@198.112.75.205) has joined #apache-wink 01[08:13] Hi Dan 01[08:13] Jason, I don't have concrete numbers on performance, but we didn't see anything that scared us. [08:14] k [08:14] we will probably want to add some more caching code that we can discuss on the mailing list (hard to do in one line chats for me) [08:14] dkulp, context so far - http://slexy.org/view/s21sCQL4NW 01[08:15] So, before we dive a little further down... any other questions/comments on the HP code as the base? [08:15] We ran some initial performance testing of the HP runtime and it seems that it performs as well as Jersy and much better then RESTEasy and Restlet 01[08:16] nadav, do you remember what kind of work the test was doing? (simple GETs, combos with POST/DELETE, etc?) [08:17] simple GET for a simple string, XML JAXB and Atom 01[08:17] Cool, thanks. 01[08:17] Again, I apologize for starting us off by diverging from the agenda. 01[08:18] I'll try to get us back there. 01[08:18] To summarize the previous discussion though: an email should be send to the list calling for a vote on the source base. Any volunteers? [08:19] i will do it 01[08:19] Thanks Martin 01[08:19] Looking at the agenda, the first item on the list is short term goals. 01[08:20] Beyond what's listed, the shortest term goal I see is to get the source base seeded in trunk. 01[08:20] If the vote to use the HP code goes through (72 hour wait at the shortest) 01[08:21] that would mean we could seed the repo by... Friday, is that right? [08:21] right [08:22] the next step would be to set a build environment so we could generate the first build 01[08:22] after seeding the source? or before? [08:23] the first build would be after we seed the code, but ..\ [08:24] I think we can start start setting the environment to save time [08:25] Do we have access to the build server? how does it work? 01[08:25] so, by setting the environment, do you mean defining a set of maven modules? 01[08:25] ah... you mean the automated build stuff? [08:25] yes 01[08:25] Ah, ok... I think Jason would be the person that knows best there. 01[08:26] Specifically, that's related to the question on the list the other day about Continuum vs. Hudson. 03[08:26] * dkulp_ (n=dkulp@host-131-239-31-200.customer.veroxity.net) has joined #apache-wink [08:26] in order to create a build you need only run mvn command 01[08:26] Once we have code in trunk that uses a maven build, we can setup those automated processes to run and dump nightly builds somewhere. 01[08:26] Jason, do you know what needs to be done to get Hudson running? [08:27] folks, feel free to start working in the sandbox till the vote is final - https://svn.apache.org/repos/asf/incubator/wink/sandbox/ 01[08:28] thanks dims [08:28] i think Jason stepped out (idle for a while) 01[08:28] ok, we can come back to that... [08:28] If he cannot setup hudson, I can. [08:28] so start with the hp base, add on ibm test cases and get the mvn build to work 01[08:28] I'll make a note of the action for build automation and follow-up. 01[08:28] thanks Dan! [08:28] then dkulp and jason can help run it on hudson 01[08:28] dims, yep, that makes sense. [08:29] sorry was outside for a second 01[08:29] no worries... were just talking about build automation for a sec. 01[08:29] if you don't mind though, we'll come back to that later. [08:30] yes, I have company atm, have to attend to them, will be back in a few 01[08:30] so far we have: short term goal of a) getting the vote public and b) getting the repo seeded. [08:30] next, So infrastructure wise what else do you guys need? 01[08:30] can we talk about next steps beyond that? 01[08:30] hmm... 06[08:30] * dims needs to get id for talial... 01[08:31] thanks for mentioning that... 01[08:31] so, we have a repo, we will have the automated build 01[08:31] do we need to do anything special to get the snapshots published? [08:31] Not from hudson 01[08:31] cool [08:32] From anyone else, yes, but not from Hudson. :-) 01[08:32] hehe... 01[08:32] Dims, I can't think of anything else infrastructure wise. 01[08:32] Oh, the TCK, but Dan mentioned the other day that they already have it up and running. [08:33] Yep. CXF passes it. 01[08:33] is that 1.0 or 1.1 [08:33] 1.0, we've put in a request for 1.1, but 1.1 isn't final yet. [08:33] Thus, not sure we'll get it until it's final. 01[08:33] makes sense... 01[08:34] so... that brings me to the next point on the agenda in the way of short term goals. 01[08:34] JAX-RS 1.0 compliance [08:34] do we need permissions to the apache mvn repository ? 01[08:34] permissions to access? no, shouldn't 01[08:34] permissions to publish to it? not sure [08:35] Eventually, for "release", it will depend on if you plan to release via Nexus or not. [08:35] Some projects are moving to Nexus (which requires permissions), others, like CXF, have not (yet). [08:35] elib, you can upload to apache mvn repository either by copying files (scp) to people.apache.org or using nextus [08:35] For those cases, the perms on people.apache.org are OK. [08:36] right, just need to make sure everyone's id is in the incubator unix group 01[08:36] dims/dan, do you have a suggestion as to which tool we should use? [08:36] stick to scp for now [08:36] If the project DOES decide to use Nexus, that will require Brian Fox to setup some stuff. [08:36] right [08:36] let's get our feet wet 01[08:36] k, thanks [08:37] before i forget, i need a volunteer to help me with updating incubator documentation (example - http://incubator.apache.org/projects/) 01[08:37] board reports and such? [08:37] and getting ready for our first board report [08:37] yep 01[08:37] I'm happy to do it. [08:37] when is that due? [08:37] Last week. [08:37] typically 3rd week of the month [08:37] Missed it. :-) [08:37] great. :) 01[08:38] do we need to put one out now, or just wait for the next? [08:38] Just wait for the next. 01[08:38] k [08:38] ngallardo, start with creating a wink.xml (https://svn.apache.org/repos/asf/incubator/public/trunk/site-author/projects/) 03[08:38] * ckmason (n=chatzill@32.97.110.55) has joined #apache-wink 02[08:38] * ckmason (n=chatzill@32.97.110.55) Quit (Client Quit) 01[08:38] thanks dims 01[08:38] all, anything else from an infrastructure perspective? [08:38] i'll get the web site updated with that. 03[08:38] * ckmason (n=chatzill@32.97.110.55) has joined #apache-wink 01[08:39] ok, I'll take that as a no. [08:39] for the board report, please start a page on the wiki 01[08:40] will do... 02[08:40] * dkulp (n=dkulp@198.112.75.205) Quit (Read error: 110 (Connection timed out)) [08:40] i also need to figure out how to generate a wink web site from our wiki contents 01[08:40] our wiki, or the boards? [08:40] our wiki 01[08:40] k 01[08:40] ah, so that we have something that ends up under incubator.apache.org/wink? [08:40] then we copy it over to incubator wiki and the pmc chair sends it to the board [08:40] ngallardo, yep 01[08:40] great 01[08:41] going back to goals, can we agree that for an initial pass JAX-RS 1.0 compliance is one of the main goals? [08:41] +1 [08:41] yes 01[08:42] excellent 01[08:42] what else would people like to see in an initial 0.1 release? 01[08:43] some of this could be extensions of discussions started on the agenda page. 01[08:43] modularity? if so, how granular? [08:43] regarding the site, the "mvn site" build a nice site with a lot of info we can use it and just replace the HP icons with Apache [08:44] back [08:44] what do you mean by modularity? 03[08:45] * dkulp_ is now known as dkulp 01[08:45] to what degree do we want to separate things like the providers from the runtime? 01[08:45] i.e., create a lean core that only pulls in what it wants. 01[08:46] or... to what degree does that already exist 01[08:46] this is where we'll need your expertise with the code 01[08:46] additionally, there's modularity for the lifecycle stuff 01[08:47] can we componetize the spring stuff to the point to where we're able to only pull that in when needed? [08:47] the Spring is a seperate module in a different jar 01[08:47] I'm trying to throw as much as I can out here to see what you guys think are the hot topics beyond spec compliance. 01[08:47] ok, good... [08:48] currently we have 3 main modules": [08:48] common, server and client, where common is used by both server and client [08:48] common+client have no servlet dependencies 01[08:48] good [08:48] and some additional modules like spring-support and webdav support 01[08:49] does the server module rely on spring explicitly? or can I run a JAX-RS endpoint without it? [08:49] providers are located in both common and server (sever contains only providers that cannot be used by client) [08:49] server does not rely on spring, it's the other way around [08:50] the spring module relies on the server 01[08:50] ok 01[08:50] but does the server module have code to deploy an endpoint without spring? [08:51] if you are asking wether the runtime can be deployed without a servlet container, then the answer is no. [08:52] the spring module also requires deployment in a servlet container [08:52] should be possible to wire up components w/o spring I would imagine, unless (forget its name, the post setting properties interface is implement) [08:52] but easy enough to use annotations for that, or configure a custom init() [08:53] yes, there is no need in Spring to run in a servlet container [08:53] IMO spring is a great container, as long as components don't implement spring-specific muck, then its super easy to wire stuff by hand, while maintaining a clean component-based architecture [08:53] its just those pesky spring-only bits you have to watch out for 01[08:54] hmm... so I think my understanding of how spring was being used is off. 06[08:54] * dims can't resist asking if there's any osgi support [08:54] not currently [08:54] what? osgi, what is this black magic you speak of? [08:54] LOL! [08:54] martinsn, ok thx. [08:54] JAX-RS and DOSGi: http://sberyozkin.blogspot.com/2009/06/distributed-osgi-ri-gets-restful-edge.html [08:55] thats easy enough to do as well 01[08:55] tying this back to the discussion of goals, I'd like to propose that one requirement for the 0.1 release be the ability to run in a web container, but without spring. [08:55] agreed [08:55] actually this is how it works now :) [08:55] sweet so 0.1 is done, nice work folks 01[08:55] ok, then I've officially confused myself... [08:55] we should also try to follow the code guidelines as much as possible i think for the 0.1 release 01[08:55] heh... [08:55] hp folks, have you also looked at existing jaxrs code in CXF? anything we can leverage from there? 01[08:56] or to that end, anything that you've seen in the IBM code that you'd like to leverage? [08:56] :) [08:56] maybe we can pul some code from some of the providers, specifically the JAXB provider [08:57] not including the testsuite na... we are going to use that (per talk before) right? 01[08:57] right, the IBM test suite will be ported over [08:57] k, cool [08:57] in addition I liked the classloading staff, that a configurable classloader is used to load the class in reflection 01[08:58] for the metadata creation stuff? [08:59] resources/providers I guess 01[08:59] ah, ok... the stuff for finding the right JAXB classes. 01[08:59] yeah, some of that comes from our experience with JAX-WS and the way JAXB works in multithreaded environments. 01[08:59] good, good... [09:00] getting back to the site discussion, running mvn site generate a very rich site with a lot of useful info about the runtime, we can use it as the wink site, we just need to replace the HP icons with Apache 01[09:00] so, we've quickly run through the hour. I'm happy to stick around, but I know it's late for the team in Israel and for Jason as well. 01[09:00] excellent... 01[09:00] I'm all for seeding the docs and adding on. 01[09:01] for the HP, have you looked at the CXF source much? [09:02] yes 01[09:02] any thoughts on concepts you'd like to borrow from there? [09:02] am sure dkulp can help with getting us smaller jars of their stuff for us to use 03[09:02] * jaydm (n=quassel@198.pubint.com) has joined #apache-wink [09:03] the integration providers looks like a very nice concept 01[09:03] integration providers? as in Java EE integration? [09:04] there is the Inte [09:04] oops [09:04] I mean the IntegrationRegistry class stuff 01[09:05] ah, agreed. a general registry for integration points/plug-ins would be helpful. 01[09:05] on the list for 0.1? [09:05] about the site, I would recommend using confluence for the main site and then suplimenting with the mvn generated bits [09:06] I don;t think it's something we want for 0.1. I think we want to get to 0.1 as fast as we can [09:06] jdillon, +1 01[09:07] nadav, +1 to 0.1 as quickly as possible. 01[09:07] to that end, any voluteers to move the code over to the sandbox and get everything working while we wait for the vote? [09:08] I'll do it 01[09:08] one thing that needs to be done there is renaming the packages to a suitable org.apache. [09:08] "svn cp" should be enough 01[09:08] thanks Michael [09:08] I am available to help with any mvn issues, though as I'm not a commiter it will be patch hell for someone [09:08] reminds me....jdillon and rott are itching to get started...both are existing apache committers [09:09] I think it should be org.apache.wink.* :) 01[09:09] are all of the dependencies being pulled down from mvn repos? 01[09:09] elman, +1 [09:09] elman: yes that looks about right to me [09:09] yes, everything from repos 01[09:09] perfect [09:09] ngallardo: afaik yes [09:09] built just fine 01[09:10] so, we've gone through a number of things on the agenda. 01[09:11] do we want to talk about things we haven't touched on yet? (third party dependencies, plug points, etc.) 01[09:11] or, have folks had enough and do you want to move the discussion to the list? [09:11] third party deps is a good discussion topic. [09:12] wanna officially wrap up the "formal" portion of the chat? [09:12] thirdparty as in? what stuff is sucked up into the distro? [09:12] folks can hang around and talk 01[09:12] dims, sure... [09:13] ngallardo, please post a log to the wink-dev mailing list 01[09:13] I'll post a note to the list with the actions discussed as well [09:13] sounds good. 01[09:13] jdillon, was actually meaning more the ATOM and JSON support. 01[09:13] someone added an agenda item to discuss this. [09:14] ok 01[09:14] hp team, this is all currently supported from within the runtime, right? [09:14] yes, ATOM is internally based on JAXB, and JSON is based on internal classes as well 01[09:15] good, I like the ATOM over JAXB route. 01[09:15] do you know of any compatibility issues with Abdera? [09:16] what dou you mean by compatibility? 01[09:17] I'm assuming that everything generated is standard... or, have you experienced any problems with Abdera consuming what you're producing... or vice/versa. [09:17] we haven't tested against abdera, but everything was implemented according to the Atom spec so ther e shoudln't be any problems 01[09:19] excellent... 01[09:19] so, I don't know what was intended for the discussion on the agenda. 01[09:19] but, the IBM code currently uses Abdrea 01[09:19] Abdera 01[09:19] I'm fine going with the interal ATOM over JAXB impl though. 01[09:21] other third party deps... the IBM code uses Jettison for JSON support. 01[09:21] can the HP code currently support a similar model of creating JSON from a JAXB object? [09:22] HP runtime uses the classes from json.org internally, not from a 3rd party module [09:22] and yes, HP has a JAXB to JSON provider [09:22] ngallardo: http://twitter.com/dejanb/statuses/2291685750 01[09:23] dkulp, interesting... where is the library located? Codehaus? Apache? [09:23] It's Tatu so probably codehaus..... [09:23] http://jackson.codehaus.org/ [09:24] dkulp, could you please guide us as a mentor? :) 01[09:24] interesting... 02[09:24] * martinsn (i=c0762d02@gateway/web/freenode/x-6182aace87ed39e6) Quit (Ping timeout: 180 seconds) 03[09:25] * martinsn (i=d47589c1@gateway/web/freenode/x-da053d7ab9873533) has joined #apache-wink 01[09:25] ok folks, it sounds like we have a clear direction in the short term. 01[09:26] I believe martin is going to post a vote to use the HP code [09:26] I have one thing I'd like to interject, that being the svn structure, I think it would serve the wink community to start out in a sub-project structure, ie no trunk/tags/branches at the toplevel of incubator/wink, 01[09:26] and in the mean time, elman is going to get the HP code seeded under the sandbox [09:26] just abt to [09:26] but thats just my anal structure talking 01[09:27] so, how will we mark point releases out of the repo? [09:27] um, I mean to start out with incubator/wink/wink-core/{trunk/tags/branches} [09:27] to allow for peir trees in symetry [09:27] as we do in geronimo [09:28] not a big deal though really [09:28] but for example the testsuite might be its own seperate project as a peer to wink-core [09:28] does that mean that each module will have a seperate lifecycle? [09:28] moving the trunk/tags/branches namespace up one helps with that sym [09:28] no [09:29] consider this: https://svn.apache.org/repos/asf/geronimo/ [09:29] we have the main server project, the gshell project, the samples, blah, blah 01[09:30] so they each release separately? [09:30] jason, I'd rather see a structure more similar to axis2 [09:30] each sub-project yes [09:30] modules, rather that sub projects. [09:30] maybe I'm not understanding.. 01[09:30] jdillon, I see what you're saying. [09:31] its fine to start, though every project I've worked with so far has had the need to have sub-projects 01[09:31] I'm wondering if we'll really have enough independent components to justify. [09:31] so rather than force an svn mv later, I would rather see that earlier [09:31] ngallardo, that's what I was wondering too.. [09:31] sure, well lets see... I'm just giving you my advise :-) 01[09:31] mind if we revist as we start tearing it apart? [09:32] its easy enough to move stuff around later 01[09:32] absolutely... I appreciate it, and hope you'll continue to be vocal. [09:32] sure, np... again just my advise 01[09:32] I'd encourage all to not be short on opinions either... 01[09:32] we need that kind of discussion 01[09:33] ok, so we've kept you here for 1.5 hours... some people want to go home, I'm sure, others want to go to sleep. [09:33] opinons are like assholes... everyones got one :-P [09:33] +1 01[09:33] unless there's anything else pressing, I say we move discussions to the list and go from there. [09:33] I think it was a good discussion and a good start for the wink community [09:34] +1 01[09:34] absolutely... great participation. thanks to everyone for joining and commenting. [09:35] thanks everyone [09:36] good morning/night everyone 02[09:36] * elman (i=c0762d02@gateway/web/freenode/x-5815cbdd2c87f316) Quit ("cya") 03[09:37] * martinsn (i=d47589c1@gateway/web/freenode/x-da053d7ab9873533) has left #apache-wink [09:37] thanks everyone, bye 02[09:38] * elib (i=c0762d02@gateway/web/freenode/x-ba414dc9475da041) Quit [09:38] bye all, and thanks everyone 02[09:38] * nfischer (i=c0762d02@gateway/web/freenode/x-761722173e3d79d0) Quit 02[09:45] * jdillon (n=jdillon@58.147.49.227) Quit 02[10:50] * gtruty (n=gregtrut@32.97.110.64) Quit 03[11:04] * dkulp_ (n=dkulp@198.112.75.205) has joined #apache-wink 02[11:05] * dkulp (n=dkulp@host-131-239-31-200.customer.veroxity.net) Quit (Read error: 60 (Operation timed out)) 03[11:20] * dkulp__ (n=dkulp@host-131-239-31-200.customer.veroxity.net) has joined #apache-wink 03[11:26] * jaydm (n=quassel@198.pubint.com) has left #apache-wink ("Later!") 02[11:35] * dkulp_ (n=dkulp@198.112.75.205) Quit (Read error: 110 (Connection timed out)) 03[11:54] * dkulp (n=dkulp@198.112.75.205) has joined #apache-wink 02[11:59] * dkulp__ (n=dkulp@host-131-239-31-200.customer.veroxity.net) Quit (Read error: 104 (Connection reset by peer)) 03[12:02] * dkulp_ (n=dkulp@host-131-239-31-200.customer.veroxity.net) has joined #apache-wink 02[12:16] * dkulp (n=dkulp@198.112.75.205) Quit (Read error: 110 (Connection timed out)) 03[13:59] * ckmason_ (n=chatzill@32.97.110.64) has joined #apache-wink 02[14:17] * ckmason (n=chatzill@32.97.110.55) Quit (Read error: 110 (Connection timed out)) 03[14:47] * ckmason (n=chatzill@32.97.110.64) has joined #apache-wink 02[14:56] * ckmason_ (n=chatzill@32.97.110.64) Quit (Read error: 110 (Connection timed out)) 03[15:00] * ckmason_ (n=chatzill@32.97.110.55) has joined #apache-wink 02[15:05] * ckmason (n=chatzill@32.97.110.64) Quit (Read error: 60 (Operation timed out)) 03[15:05] * ckmason_ is now known as ckmason Session Close: Tue Jun 23 15:19:55 2009 Session Start: Tue Jun 23 15:19:55 2009 Session Ident: #apache-wink 02[15:19] * Disconnected 02[15:21] * Attempting to rejoin channel #apache-wink 03[15:21] * Rejoined channel #apache-wink 02[15:36] * ngallardo (n=nlgallar@32.97.110.64) Quit (Read error: 110 (Connection timed out))