Differences between revisions 4 and 5
Revision 4 as of 2004-12-08 15:14:23
Size: 31989
Revision 5 as of 2009-09-20 22:47:17
Size: 31989
Editor: localhost
Comment: converted to 1.6 markup
No differences found!

topics discussed: ServiceDesc, relation to WSDL, WOM, where to hang user properties in engine and WSDL (if at all), and NAMINGConvetions:


<sanjiva> ok are we converging? shall I try to summarize (at the risk of getting shot)
<sanjiva> (1) we create org.apache.wsdl.* to be the WSDL object model with an extensibility hook to store arbitrary runtime stuff via a hashmap
<sanjiva> (2) we create o.a.axis.description.AxisOperation etc. classes that extend the corresponding class in o.a.wsdl and provides a nice API to get/set the stuff we care about for each WSDL component
<sanjiva> (3) when a service is deployed the WSDLService structure is created with all the additional properties etc. 
<sanjiva> (4) The registry stores AxisService etc. 
<sanjiva> (5) Alek raises a minority objection ;-)
<sanjiva> (6) The Module stuff is still independent .. we'lll prolly still need a ModuleDesc type thing to store the module data in memory. Need to come up with a consistent naming scheme to make them fit together nicely.
<alek_s> Let be noted that i am opposed to use of (Hash)Map  but i am OK for now ...
<gdaniels> that sounds good, but of course we need to see the actual design

<gdaniels> I need to take off soon to attend some meeting-related stuff downstairs.  Would love to discuss the engine next week.
<gdaniels> Handlers, phases, etc

<gdaniels> chat.setProperty("minorityObjection", aleksThing)
(and we know from discussion that this particualr property may be lost during serialization ;-) )


[12/8/2004 8:56 AM] =-= Topic for #apache-axis is ``Apache Axis Web service Framework - WSDL WSDD SOAP XMLRPC''
[12/8/2004 8:56 AM] =-= Topic for #apache-axis was set by FR^2 on Monday, December 06, 2004 6:58:53 AM
[12/8/2004 8:59 AM] <gdaniels> good $TIME_OF_DAY, folks!
[12/8/2004 9:00 AM] <alek_s> good $DAY_OR_NIGHT :)
[12/8/2004 9:00 AM] <Srinath> Hi glen 
[12/8/2004 9:00 AM] <Deepal> Hi all
[12/8/2004 9:00 AM] <Ajith> hi all
[12/8/2004 9:00 AM] <Srinath> good morning/night :)
[12/8/2004 9:01 AM] <Harsha> Hi All
[12/8/2004 9:01 AM] <Srinath> what is sheduled for today?
[12/8/2004 9:01 AM] <Srinath> M1 ?
[12/8/2004 9:01 AM] =-= gdaniels has changed the topic to ``Weekly Axis2 chat''
[12/8/2004 9:01 AM] <gdaniels> Hey, why isn't the chat info on the front page of the Wiki anymore?
[12/8/2004 9:02 AM] <Ajith> WSDL and deployment
[12/8/2004 9:02 AM] <Srinath> let me see what happen to front page
[12/8/2004 9:02 AM] <Ajith> it is !
[12/8/2004 9:02 AM] <Ajith> aaah the link is missing :)
[12/8/2004 9:02 AM] -->| chathura (~chathura@ has joined #apache-axis
[12/8/2004 9:02 AM] <Ajith> BTW very small crowd
[12/8/2004 9:02 AM] <Ajith> Hi Chathura
[12/8/2004 9:02 AM] <chathura> hi all
[12/8/2004 9:03 AM] <chathura> hi ajith
[12/8/2004 9:03 AM] <Srinath> the chat info is at  http://wiki.apache.org/ws/Axis2/People
[12/8/2004 9:03 AM] <gdaniels> Small crowd often means good ability to get things done :)
[12/8/2004 9:03 AM] <gdaniels> Why is it there?
[12/8/2004 9:03 AM] <Ajith> :)
[12/8/2004 9:03 AM] <chathura> :)
[12/8/2004 9:03 AM] <Deepal> :)
[12/8/2004 9:03 AM] <gdaniels> It's not about people, it's about the development process, should definitely be on the first page IMO.
[12/8/2004 9:04 AM] <Srinath> will fix it 
[12/8/2004 9:04 AM] <alek_s> http://wiki.apache.org/ws/ChatAgenda/20041208
[12/8/2004 9:05 AM] <gdaniels> I'd like to talk about the ServiceDesc thread
[12/8/2004 9:05 AM] <Srinath> me too 
[12/8/2004 9:05 AM] <Srinath> there is a mail regarding that 
[12/8/2004 9:05 AM] <chathura> +1:)
[12/8/2004 9:05 AM] <Srinath> let me find subject
[12/8/2004 9:05 AM] <gdaniels> yes, hence "thread"
[12/8/2004 9:06 AM] <gdaniels> Using the WSDLService in the registry[continuingRE: [Axis2][Engine]Step by Step;Engine, Engine registry deployment and the Phase Resolver]
[12/8/2004 9:06 AM] <Srinath> mail subjectr:Using the WSDLService in the registry[continuingRE: [Axis2][Engine]Step by Step;Engine, Engine registry deployment and the Phase Resolver]
[12/8/2004 9:06 AM] <gdaniels> :)
[12/8/2004 9:06 AM] -->| sanjiva (~sanjiva@ has joined #apache-axis
[12/8/2004 9:06 AM] <Srinath> yes glen :)
[12/8/2004 9:06 AM] <sanjiva> hi yall sorry to be late
[12/8/2004 9:06 AM] <Harsha> Hi Sanjiva
[12/8/2004 9:07 AM] <gdaniels> That's OK, Sanjiva, we assigned you the AI of finishing the code for us by next week.
[12/8/2004 9:07 AM] <gdaniels> Hope that's OK! :)
[12/8/2004 9:07 AM] <sanjiva> ah sure! :-)
[12/8/2004 9:07 AM] <Srinath> +1
[12/8/2004 9:07 AM] <Srinath> :D
[12/8/2004 9:07 AM] <gdaniels> So we're about to discuss ServiceDesc, it seems 
[12/8/2004 9:07 AM] <Deepal> yep.
[12/8/2004 9:07 AM] <gdaniels> At the F2F we talked about using the WSDL structure to hang our metadata off of.
[12/8/2004 9:07 AM] <Ajith> chathura?
[12/8/2004 9:08 AM] <gdaniels> I'm wondering why we wouldn't want to do that still.
[12/8/2004 9:08 AM] <gdaniels> (I used to hate that idea, but I'm liking it more and more)
[12/8/2004 9:08 AM] <Ajith> mmm wouldn't that corrupt the WOM
[12/8/2004 9:08 AM] <Ajith> ?
[12/8/2004 9:08 AM] <gdaniels> not at all
[12/8/2004 9:08 AM] <chathura> no ajith
[12/8/2004 9:08 AM] <chathura> the reason was 
[12/8/2004 9:08 AM] <Srinath> glen that means there is no metadata .. and we populate info to WSDLService?
[12/8/2004 9:08 AM] <gdaniels> WSDL is extensible, first of all.  Also, the WSDL model we use should be even more extensible (stuff can hang off it that isn't necessarily serialized)
[12/8/2004 9:08 AM] <sanjiva> Ah yes .. let's talk about that .. I think we are still doing that .. so let's assume we will introduce a "userData" property to every type of node in the WSDL component model
[12/8/2004 9:08 AM] <chathura> we wanted give a better api for the engine programmer
[12/8/2004 9:09 AM] <sanjiva> let that be a hashtable
[12/8/2004 9:09 AM] <gdaniels> Srinath - of course there's metadata!
[12/8/2004 9:09 AM] <gdaniels> We love metadata
[12/8/2004 9:09 AM] <gdaniels> It's just where it lives, and on what kind of objects
[12/8/2004 9:09 AM] <sanjiva> then we can put whatever stuff we want .. like per operation specific modules or whatever
[12/8/2004 9:09 AM] <gdaniels> +1 Sanjiva
[12/8/2004 9:09 AM] <gdaniels> That's the general idea
[12/8/2004 9:10 AM] <Srinath> +1
[12/8/2004 9:10 AM] <gdaniels> then if there are specific things we want to bake in, we can put those in outside the userData part
[12/8/2004 9:10 AM] <sanjiva> then the only "ugliness" is that to get say the inFlowChain for an operation, you'd need to do:
[12/8/2004 9:10 AM] <gdaniels> if you look at the 1.X ServiceDesc classes, a lot of that info is the exact same info you need in a WSDL model
[12/8/2004 9:10 AM] <sanjiva> service.operation.getUserData().get("inFlowChain") ..
[12/8/2004 9:10 AM] <sanjiva> the only reason I'm for keeping the ServiceDesc class is to make that call look like OperationDesc.getInFlowChain ()
[12/8/2004 9:10 AM] <gdaniels> Sanjiva: we could also do this:
[12/8/2004 9:11 AM] <gdaniels> class AxisOperation extends WSDLOperation {
[12/8/2004 9:11 AM] <sanjiva> undernearth it should still be doing that call .. going to the WSDL model and pulling the data out
[12/8/2004 9:11 AM] <gdaniels>   public InFlowChain getInFlowChain();
[12/8/2004 9:11 AM] <gdaniels> }
[12/8/2004 9:11 AM] <gdaniels> etc
[12/8/2004 9:11 AM] <sanjiva> ah that's a better plan!
[12/8/2004 9:11 AM] <gdaniels> lots of ways to skin this cat
[12/8/2004 9:11 AM] <sanjiva> So instead of OperationDesc we have AxisOperation .. I like it!
[12/8/2004 9:12 AM] <sanjiva> we'd still need to cast the type I guess to get to AxisOperation (as Interface.getOperation(qname) will return an Operation) but that's ok
[12/8/2004 9:12 AM] <Srinath> yes the additional info can be stroed like e.g. HashTable properties .. but we can wrap can give nice API by ServiceDesc
[12/8/2004 9:12 AM] <gdaniels> we need to think about what kinds of stuff we put in there, and how/if it affects the WSDL serialization, but I think it can work
[12/8/2004 9:12 AM] <Srinath> 1)Handlers 2)properties
[12/8/2004 9:12 AM] <Srinath> ect ..
[12/8/2004 9:13 AM] <chathura> so this is basically a wrapper am i right
[12/8/2004 9:13 AM] <gdaniels> Well here's a question.  Where does the info about the back-end implementation class go?
[12/8/2004 9:13 AM] <chathura> a wrapper that will wrap the component model
[12/8/2004 9:13 AM] <gdaniels> chathura: Yup
[12/8/2004 9:13 AM] <Deepal> Handler mean not only the executable handler ?
[12/8/2004 9:13 AM] <sanjiva> Yes at one point I was proposing that we use the WSDL syntax itself instead of our own service.wsdd, but I changed my mind after convincing myself that the WSDL syntax would be very painful for users
[12/8/2004 9:13 AM] <Deepal> it should also keep pahse rules etc...
[12/8/2004 9:14 AM] <sanjiva> Glen: to me the impl class stuff is an endpoint .. so it goes in the Endpoint object
[12/8/2004 9:14 AM] <alek_s> Just a comment: i think  it would be good to consider more type safe to cast metadata into Java interfaces
[12/8/2004 9:14 AM] <Srinath> glen e.g. that back end info go in to the WSDLService as a property .. but it is nicley wrapped by ServiceDesc
[12/8/2004 9:14 AM] <gdaniels> I think it's more appropriate in the service, actually, snajiva
[12/8/2004 9:14 AM] <gdaniels> s/snajiva/sanjiva/
[12/8/2004 9:15 AM] <gdaniels> different endpoints are just different ways to get to the "same thing"
[12/8/2004 9:15 AM] <sanjiva> Hmmm. Maybe I snafooed. Hmm. But wait, that's not consistent with soap endpoints for example ..
[12/8/2004 9:15 AM] <alek_s> something like: FlowChainMetada fcmd = (FlowChainMetada) service.operation.getUserDataCast(FlowChainMetada.class); fcmd.doWhatever()
[12/8/2004 9:16 AM] <Deepal> alke : if you are refer to excution cahin 
[12/8/2004 9:16 AM] <chathura> hmm my idea was it should go into the service ....
[12/8/2004 9:17 AM] <Deepal> we will do somthing smilar to thet
[12/8/2004 9:17 AM] <Deepal> that
[12/8/2004 9:17 AM] <alek_s> i do nto like hashtables annd mahoc strings constants - it creates second kind of API that is very fragile 
[12/8/2004 9:18 AM] <gdaniels> alek: I don't agree that it's fragile
[12/8/2004 9:18 AM] <sanjiva> So now this is going back to what I was proposing back at the F2F ... treat a service as a WSDL binding (ala WSIF). 
[12/8/2004 9:18 AM] <gdaniels> I think that kind of thing is VERY useful for loosely coupled and extensible portions of a system
[12/8/2004 9:18 AM] <sanjiva> Alek: if the string key for the hashtable is burnt into a subclass its not fragile right?
[12/8/2004 9:18 AM] <Srinath> glemn, Sir can u explained bit more on endpoint? 
[12/8/2004 9:18 AM] <Srinath> sorry glen
[12/8/2004 9:18 AM] <sanjiva> let's finish the fragility thread first :)
[12/8/2004 9:18 AM] <alek_s> Glen: it is fragile as it is not checked during compiel time
[12/8/2004 9:19 AM] <sanjiva> things get really fragile if we start having multiple threads
[12/8/2004 9:19 AM] <alek_s> Sanjiva: yes - you can add wrappers and they hide problems 
[12/8/2004 9:19 AM] <sanjiva> Alek: How do you make it compile time checked? The WSDL model is a constant .. we can't put Axis2 specific stuff in there; all we can put are hooks
[12/8/2004 9:20 AM] <Srinath> or properties
[12/8/2004 9:20 AM] <gdaniels> alek: Extensibility means to some extent a lack of compile time checking.  I think that's OK.  You can mitigate it by providing a casting API, but I tend not to love those.
[12/8/2004 9:21 AM] <sanjiva> Srinath: ??
[12/8/2004 9:21 AM] <alek_s> Glen: i am more concerned about having it somewhat under control 
[12/8/2004 9:21 AM] <Srinath> means hooks = properties
[12/8/2004 9:22 AM] <alek_s> Glen: and i am talking abiut dynamic casts so it is also extensible ...
[12/8/2004 9:22 AM] <chathura> well this is how i feel there are things that we shouldnt change WSDL component based thing we did so i doubt we can make it compile time checked at all unless
[12/8/2004 9:22 AM] <chathura> we come up witht a parellel object hierarchy
[12/8/2004 9:22 AM] <gdaniels> Hm.  I think we need to ground this discussion in specific examples to make sure we aren't talking around each other.
[12/8/2004 9:22 AM] <Srinath> sir, by "or properties" I means hooks=propertis
[12/8/2004 9:23 AM] <alek_s> Wouldnt it be better to have XML tree as place to hang user data for extensions?
[12/8/2004 9:23 AM] <gdaniels> Alek, let's say I want to add a property at runtime to a service, like I can do today with service.setOption()
[12/8/2004 9:24 AM] <gdaniels> Do you think we should take out the setOption(String, Object)/getOption() APIs?
[12/8/2004 9:24 AM] <gdaniels> I would be STRONGLY -1 to that
[12/8/2004 9:24 AM] <gdaniels> msgContext.setProperty/getProperty is one of the major design strengths of Axis, IMO
[12/8/2004 9:25 AM] <Srinath>  glen:yes the setOptions shuould be there !
[12/8/2004 9:25 AM] <alek_s> i would add namespace to property name to avoid clashes as minimum
[12/8/2004 9:25 AM] <alek_s> setOption(XmlNamesapce ns, String name, Object value)
[12/8/2004 9:25 AM] <gdaniels> Figuring out what should be type-checked actual APIs and what should be generic Map-like APIs is an art, not a science
[12/8/2004 9:25 AM] <gdaniels> ew
[12/8/2004 9:25 AM] <Srinath> properties are strings is not they?
[12/8/2004 9:25 AM] <alek_s> Glen: this is an oppurtinity to try to make it better ...
[12/8/2004 9:27 AM] <Srinath> alek_s: what is the better option? I mean we have to choose one at the end 
[12/8/2004 9:27 AM] <alek_s> i think Glen is talking about WSDL proeprties not Java properties
[12/8/2004 9:27 AM] <gdaniels> I'm talking about Java properties, I think...
[12/8/2004 9:27 AM] <gdaniels> i.e. not stuff that necessarily gets written into WSDL xml
[12/8/2004 9:27 AM] <sanjiva> wait, let's make it more concrete .. 
[12/8/2004 9:27 AM] <gdaniels> stuff that exists at runtime
[12/8/2004 9:27 AM] <Srinath> I think wsdl properties become java propertis 
[12/8/2004 9:27 AM] <alek_s> key quesiton is how we document those properties in API and then enforce that right one are used - they are like GOTO - they can show up anywhere ...
[12/8/2004 9:27 AM] <gdaniels> Yes, Srinath
[12/8/2004 9:28 AM] <sanjiva> org.apache.wsdl.WSDLOperation has a method called setProperty (string, object) and a corresponding getProperty(); yes?
[12/8/2004 9:28 AM] <Srinath> +1
[12/8/2004 9:28 AM] <gdaniels> think so
[12/8/2004 9:28 AM] <chathura> yup
[12/8/2004 9:28 AM] <sanjiva> So Alek's concern is that that's an open invitation for people to store whatever crap they want in the WSDLOperation, right Alek?
[12/8/2004 9:29 AM] <alek_s> yes and AXIS1 msgContext can contain anthing including kitchen sink (Engine :)) 
[12/8/2004 9:29 AM] <sanjiva> and that we have no control over what gets stuffed in there (nor any compile time checks)
[12/8/2004 9:29 AM] <alek_s> that makes it interestign challenge to serialzie it ...
[12/8/2004 9:29 AM] <alek_s> and for example to pause or persist handler chain 
[12/8/2004 9:29 AM] <sanjiva> Yes I agree we need to control (by documentation at least) what gets stored in message context, but let's focus on WSDLOperation for now
[12/8/2004 9:30 AM] <sanjiva> So, we have another option then - we can use WSDL's extensibility stuff .. 
[12/8/2004 9:30 AM] <Srinath> sir you mean checking the option is known one?
[12/8/2004 9:30 AM] <sanjiva> isntead of setProperty() etc., we can do "addChild (OmElement) to add a child to the WSDLOperation ..
[12/8/2004 9:30 AM] <gdaniels> ew ew ew
[12/8/2004 9:30 AM] <gdaniels> -1 to that, sanjiva
[12/8/2004 9:30 AM] <sanjiva> I know ;-) ew from me too ...
[12/8/2004 9:31 AM] <alek_s> Sanjiva: so WSDL is new object model that is different completely from XML Infoset?
[12/8/2004 9:31 AM] <alek_s> Sanjiav: +1 - that would be serializable and XML element namespaces would protet against collisions
[12/8/2004 9:31 AM] <Deepal> can we add omelemnt to wsdl ?
[12/8/2004 9:31 AM] <Srinath> where the OM element comes in :)
[12/8/2004 9:31 AM] <gdaniels> WSDL object model != XML infoset, Alek, correct.
[12/8/2004 9:32 AM] <Srinath> WOM != OM !!
[12/8/2004 9:32 AM] <sanjiva> Well the problem is if Alek (or someone) wants to serialize the WSDL object model then the random props can show up ..
[12/8/2004 9:32 AM] <gdaniels> There are two things going on here, and I think we should separate them out.
[12/8/2004 9:32 AM] <sanjiva> yes I know Srinath but (I forgot too) there can be arbitrary extensibility in a WSDL document ...
[12/8/2004 9:32 AM] <gdaniels> First, there is the idea of attaching runtime stuff to the WSDL object model.  Our own metadata, etc.
[12/8/2004 9:33 AM] <Srinath> :)
[12/8/2004 9:33 AM] <Srinath> best case we can check the property is it a know one when we set
[12/8/2004 9:33 AM] <gdaniels> This is SEPARATE from the second thing, which is about how our own extensions affect the WSDL serialization
[12/8/2004 9:33 AM] <gdaniels> I'm trying to talk about the first, and not the second, right now.
[12/8/2004 9:33 AM] <sanjiva> +1 Glen ... but do we need to worry about "serializing"/persisting the first case somehow via the WSDL OM?
[12/8/2004 9:34 AM] <sanjiva> (Yes let's not talk about the 2nd one for now .. that's the one where we need to re-eval the WSDL OM and AXIOM relationship. Maybe next week.)
[12/8/2004 9:34 AM] <gdaniels> sanjiva: I'm not sure - I could see ways of doing it where the answer is yes and other ways where the answer is no.
[12/8/2004 9:34 AM] <sanjiva> I think that's part of what concerns Alek .. if we somehow merge 1 & 2 then you get persistence free
[12/8/2004 9:34 AM] <alek_s> about persitence: if OmElement implements NotXmlSerializable it would be just skipped and essentially be only available runtime  - disappear when serializaed ...
[12/8/2004 9:35 AM] <Deepal> I think it is better if we can talk what we want for the M1  :)
[12/8/2004 9:35 AM] <alek_s> persistence is oneof fundamental concerns (liek security) it affect everything
[12/8/2004 9:35 AM] <Srinath> Deeapl +1
[12/8/2004 9:35 AM] <alek_s> and everything is easier if you do not have to worry about it
[12/8/2004 9:36 AM] <gdaniels> well there are at least two kinds of serialization too
[12/8/2004 9:36 AM] <sanjiva> well Alek let's remember that what we're talking about is the *runtime* repreesntation of something that's already persistent ...
[12/8/2004 9:36 AM] <gdaniels> there is Java serialization and then WSDL serialization
[12/8/2004 9:36 AM] <Srinath> do we want to worry about the persistance of WSDLService when read in .. are we lose something if we say it is readonly?
[12/8/2004 9:36 AM] <sanjiva> So its not really necessary to be able to store every bit of runtime state we create!
[12/8/2004 9:36 AM] <gdaniels> even if for some reason we want to serialize and restore the WSDLService to/from a database, that doesn't mean that our additional objects would change the WSDL XML....
[12/8/2004 9:36 AM] <Srinath> Sir +1
[12/8/2004 9:36 AM] <gdaniels> they might be just internal Axis things
[12/8/2004 9:36 AM] <chathura> yup
[12/8/2004 9:37 AM] <alek_s> ok i think it is nto required for M1 - but later those files may be created and modfied through API right ...
[12/8/2004 9:37 AM] <sanjiva> glen: +1
[12/8/2004 9:37 AM] <Srinath> we might go in to level that WSDLService is readonly  are we?
[12/8/2004 9:37 AM] <gdaniels> There is definitely a call to support things that DO change the WSDL serialization - but not for M1
[12/8/2004 9:37 AM] <gdaniels> Srinath: readonly -1
[12/8/2004 9:37 AM] <sanjiva> Alek: yes agreed. But we can achieve that by having the AxisOperation class do the right kind of serialization .. 
[12/8/2004 9:37 AM] <gdaniels> You should be able to generate one dynamically if you want
[12/8/2004 9:38 AM] <chathura> no think we need not provide for garenteed serialisability for now ...at least
[12/8/2004 9:38 AM] <Srinath> glen: yes I accept .. but it might make things simple .. anyway let us differ that for now:)
[12/8/2004 9:39 AM] <chathura> i was refering to the axis specific stuff btw
[12/8/2004 9:39 AM] <sanjiva> ok are we converging? shall I try to summarize (at the risk of getting shot)
[12/8/2004 9:39 AM] <gdaniels> chathura: +1
[12/8/2004 9:39 AM] <gdaniels> sanjiva: go for it
[12/8/2004 9:39 AM] <Srinath> seems we are happy to have a hashmap:)
[12/8/2004 9:40 AM] <sanjiva> (1) we create org.apache.wsdl.* to be the WSDL object model with an extensibility hook to store arbitrary runtime stuff via a hashmap
[12/8/2004 9:40 AM] <Srinath> +1
[12/8/2004 9:40 AM] <sanjiva> (2) we create o.a.axis.description.AxisOperation etc. classes that extend the corresponding class in o.a.wsdl and provides a nice API to get/set the stuff we care about for each WSDL component
[12/8/2004 9:41 AM] <gdaniels> that sounds good, but of course we need to see the actual design
[12/8/2004 9:41 AM] <alek_s> Let be noted that i am opposed to use of (Hash)Map  but i am OK for now ...
[12/8/2004 9:41 AM] <sanjiva> (3) when a service is deployed the WSDLService structure is created with all the additional properties etc. 
[12/8/2004 9:41 AM] <gdaniels> noted, Alek
[12/8/2004 9:41 AM] <gdaniels> (yay minutes :))
[12/8/2004 9:41 AM] <sanjiva> (4) The registry stores AxisService etc. 
[12/8/2004 9:41 AM] <sanjiva> (5) Alek raises a minority objection ;-)
[12/8/2004 9:42 AM] <chathura> :)
[12/8/2004 9:42 AM] <gdaniels> chat.setProperty("minorityObjection
[12/8/2004 9:42 AM] <Srinath> oh :D
[12/8/2004 9:42 AM] <gdaniels> ", aleksThing)
[12/8/2004 9:42 AM] <Deepal> :)
[12/8/2004 9:42 AM] <sanjiva> (6) The Module stuff is still independent .. we'lll prolly still need a ModuleDesc type thing to store the module data in memory. Need to come up with a consistent naming scheme to make them fit together nicely.
[12/8/2004 9:42 AM] <Srinath> seems that objection is not persistance and forgotten by next chat :)
[12/8/2004 9:43 AM] <gdaniels> LOL Srinath
[12/8/2004 9:43 AM] <sanjiva> ;-)
[12/8/2004 9:43 AM] <sanjiva> So, are we rigid enough to go back to service vs. endpoint to store the impl class etc.?
[12/8/2004 9:43 AM] <chathura> think so
[12/8/2004 9:43 AM] <Srinath> yes I think
[12/8/2004 9:44 AM] <gdaniels> Module stuff - sure, but modules have API from AxisOperation/AxisService/etc
[12/8/2004 9:44 AM] <Deepal> sir cant we keep that inside org.apache.descrption.
[12/8/2004 9:44 AM] <gdaniels> axisOperation.addModule()/getModules()/etc
[12/8/2004 9:44 AM] <sanjiva> Glen: yes, you can look them up etc. from AxisOperation etc. ..
[12/8/2004 9:44 AM] <Deepal> I mean ModuleDesc
[12/8/2004 9:44 AM] <gdaniels> +1
[12/8/2004 9:44 AM] <sanjiva> Just that you need a ModuleDesc type thing which is clearly not WSDL related 
[12/8/2004 9:44 AM] <Srinath> +1 to keep them in discription
[12/8/2004 9:44 AM] <sanjiva> Actually before moving to the Service/Endpoint discussion, let's talk about naming schemes ..
[12/8/2004 9:44 AM] <Srinath> it is a extenstion I think :)
[12/8/2004 9:45 AM] <sanjiva> WSDLOperation or WsdlOperation
[12/8/2004 9:45 AM] <Deepal> but what we are going to keep inside that pakage 
[12/8/2004 9:45 AM] <Deepal> can be use by Deployemnt module too
[12/8/2004 9:45 AM] <gdaniels> sanjiva: I prefer WSDLOperation
[12/8/2004 9:45 AM] <alek_s> as you know i am for WsdlOperation it is for me easier to read and more in JavaNaming spiriti :)
[12/8/2004 9:45 AM] <sanjiva> Alek prefers the latter .. I prefer the prior, but not hard enought to lay down on the tracks for it
[12/8/2004 9:45 AM] <gdaniels> because WSDL is the common way people refer to it
[12/8/2004 9:45 AM] <gdaniels> AxisOperation rather than AXISOperation because that's the common way people refer to it
[12/8/2004 9:45 AM] <chathura> hmm no need to refactor;)
[12/8/2004 9:45 AM] <sanjiva> So same for XMLPullParser too? What does StAX do BTW?
[12/8/2004 9:46 AM] <alek_s> StAXOMBuilderTest vs StaxOmBuilderTest
[12/8/2004 9:46 AM] <sanjiva> no that's our stuff .. I mean StAX proper
[12/8/2004 9:46 AM] <alek_s> Straming Api for XML
[12/8/2004 9:46 AM] <alek_s> but my favorite is: StAXSOAPModelBuilder  ...
[12/8/2004 9:46 AM] <Srinath> Stax is written as StAx
[12/8/2004 9:46 AM] <alek_s> it looks lie XSOAP :)
[12/8/2004 9:47 AM] <sanjiva> Alek, I mean the name of the interface .. is it XMLPullParser or XmlPullParser or something else?
[12/8/2004 9:47 AM] <alek_s> (XSOAP is my project)
[12/8/2004 9:47 AM] <sanjiva> I though XSUL was your project ;-) u have too many damned projects that's the problem!
[12/8/2004 9:47 AM] <alek_s> in stax XML convention is used
[12/8/2004 9:47 AM] <alek_s> in XPP Xml convention is used
[12/8/2004 9:47 AM] <Deepal> XSUL ?
[12/8/2004 9:48 AM] <alek_s> all classes are XsulSomething :)
[12/8/2004 9:48 AM] <Srinath> can someone decode XSUL for us :D
[12/8/2004 9:48 AM] <alek_s> XSUL is XML Services "Utility Library"
[12/8/2004 9:49 AM] <alek_s> i no longer think that everything must be SOAPized ;-)
[12/8/2004 9:49 AM] <sanjiva> So XML convention is more the Java rule right? The rule is if its an acronym you keep it that way .. IIRC
[12/8/2004 9:49 AM] <alek_s> (or RMIzed as it was SoapRMI - learning process ...)
[12/8/2004 9:50 AM] <alek_s> yes - it is prevailing convention based on assumption that people actually think about XML as acronym and not a word
[12/8/2004 9:50 AM] <Srinath> means we go for WSDL?  
[12/8/2004 9:50 AM] <sanjiva> ok shall we just take a vote on the list? We need to decide .. and its purely subjective so no point wasiting time now I guess. 
[12/8/2004 9:50 AM] <alek_s> uf you join more than two actonyms you GETREALHORRORS
[12/8/2004 9:51 AM] <sanjiva> Ah but those are not acryonims GetRealHorrors 
[12/8/2004 9:51 AM] <alek_s> it seesm it would be another mojority objection anyway ...
[12/8/2004 9:51 AM] <chathura> :)
[12/8/2004 9:51 AM] <alek_s> GET - Graphical Envorment Task etc ;-)
[12/8/2004 9:51 AM] <sanjiva> ;-) you never know Alek, there's only like 3 people voting now and Glen and I can attest to the fact that most decision in WSDL were made by the silent majority
[12/8/2004 9:52 AM] <alek_s> democracy requires participation ...
[12/8/2004 9:52 AM] <gdaniels> I need to take off soon to attend some meeting-related stuff downstairs.  Would love to discuss the engine next week.
[12/8/2004 9:52 AM] <gdaniels> Handlers, phases, etc
[12/8/2004 9:52 AM] <sanjiva> ok, so can we switch topics then again? Store impl info in WService or WEndpoint?
[12/8/2004 9:52 AM] <gdaniels> I'll put that in wiki
[12/8/2004 9:52 AM] <sanjiva> Glen: sounds good ..
[12/8/2004 9:52 AM] <gdaniels> impl info should be in service
[12/8/2004 9:52 AM] <gdaniels> imho
[12/8/2004 9:53 AM] <sanjiva> Well that's not consistent with WSDL tho Glen
[12/8/2004 9:53 AM] <gdaniels> WSDL doesn't talk about impls
[12/8/2004 9:53 AM] <sanjiva> WSDL doesn't put anything in Service .. only Endpoint
[12/8/2004 9:53 AM] <gdaniels> (does it?)
[12/8/2004 9:53 AM] <sanjiva> well endpoints are analogous to implementations .. "an endpoint is where the service is available"
[12/8/2004 9:53 AM] <gdaniels> I don't see how that matters, Sanjiva
[12/8/2004 9:53 AM] <gdaniels> We're talking about building a transport-generic service engine
[12/8/2004 9:53 AM] <sanjiva> do we want to leave room for the possibility of multiple/alternate implementations for the same service .. down the road
[12/8/2004 9:54 AM] <gdaniels> I can attach a variety of transports (endpoints) to a given service (service)
[12/8/2004 9:54 AM] <Srinath> service can stroe providers I think ..
[12/8/2004 9:54 AM] <sanjiva> glen: yes, but I'm talking about implementing the service in Java vs in XSLT, for example
[12/8/2004 9:54 AM] <Srinath> providers will allow differant impls of service 
[12/8/2004 9:54 AM] <gdaniels> sanjiva: So this is about providers
[12/8/2004 9:55 AM] <gdaniels> Srinath: yup
[12/8/2004 9:55 AM] <Srinath> :)
[12/8/2004 9:55 AM] <gdaniels> WE only have a Java impl class for certain kinds of services
[12/8/2004 9:55 AM] <gdaniels> the default kind, probably, but not all kinds
[12/8/2004 9:55 AM] <sanjiva> yes indeed .. impl and providers go together
[12/8/2004 9:55 AM] <sanjiva> yes but if we had Xalan integrated (for example) an XSLT service is an easy thing .. 
[12/8/2004 9:56 AM] <gdaniels> So a WSDLService should, IMO, tie to a single provider, and that provider has various options
[12/8/2004 9:56 AM] <sanjiva> ok. maybe that's overengineering it .. I'm ok with using Service and we can revisit if we find that to be a limitation in the future.
[12/8/2004 9:56 AM] <chathura> yup .. for m1
[12/8/2004 9:56 AM] <sanjiva> (no I meant even later future .. like Axis2 v2.x :-))
[12/8/2004 9:57 AM] <gdaniels> ok - on that note I'm gonna take off.  Enjoy, everyone!
[12/8/2004 9:57 AM] <sanjiva> ok bye 
[12/8/2004 9:57 AM] |<-- gdaniels has left irc.freenode.net ()
[12/8/2004 9:57 AM] <sanjiva> oops I have a call too in 2 mins; shall we call it a nite/morning/day?
[12/8/2004 9:57 AM] <chathura> :)
[12/8/2004 9:57 AM] <Harsha> What is the timeline for implmentation?
[12/8/2004 9:57 AM] <Srinath> Harsha dec mid/end 
[12/8/2004 9:57 AM] <sanjiva> Chathura, what's your estimate of the WOM for the basic stuff?
[12/8/2004 9:58 AM] <sanjiva> Harsha, you were looking into the WSDL stuff right? Maybe you can help with AxisOperation etc. stuff?
[12/8/2004 9:58 AM] <chathura> within next week 
[12/8/2004 9:58 AM] <chathura> i can safely say it ll be in god shape
[12/8/2004 9:58 AM] <Harsha> I can help with some development work.
[12/8/2004 9:58 AM] <Deepal> Chathura I can join with u if u want any help
[12/8/2004 9:59 AM] <Deepal> boz we need to intregtae Deployemnt and WSDL
[12/8/2004 9:59 AM] <sanjiva> How's teh WSDL reading stuf coming along? Are you implementing it using StAX or what?
[12/8/2004 9:59 AM] <Srinath> we might try to integerate WOM prototype too 
[12/8/2004 9:59 AM] <Harsha> I was looking at WSDl from a user perspective only. Don't know much on the internals!
[12/8/2004 9:59 AM] <sanjiva> yes we should put all the pieces to the proto2 dir ASAP
[12/8/2004 10:00 AM] <chathura> ok this org.apache.axis.descrption . Axis* is what i ll be working now
[12/8/2004 10:00 AM] <Deepal> Harsha : for the M1 deployment is already completed :)
[12/8/2004 10:00 AM] <chathura> wom stuff is ready to be shipped to prototype 2
[12/8/2004 10:00 AM] <Harsha> Deepal: ok:)
[12/8/2004 10:01 AM] <Srinath> Harsha I think we have areas that are not to spread in WSDL that you can work in
[12/8/2004 10:01 AM] <Srinath> cahtura what do you think .. where the gharsha can hop in?
[12/8/2004 10:01 AM] <Srinath> sorry means Harsha not g..
[12/8/2004 10:01 AM] <Harsha> Ok Let me know. Given a spec I can build something. I am nto sure how the devlopment process works here>
[12/8/2004 10:02 AM] <chathura> harsha can you look into this org.apache.axis.description thing
[12/8/2004 10:02 AM] <Harsha> Where do I start with that?
[12/8/2004 10:03 AM] <Deepal> I can provide bacis requirments from Deployment for the org.apache.axis.description
[12/8/2004 10:03 AM] <chathura> you ll have to start with the wom
[12/8/2004 10:03 AM] <Srinath> think Harsha and chathura start taling .. I will quit bye :)
[12/8/2004 10:03 AM] <Srinath> talkin
[12/8/2004 10:03 AM] <Deepal> I mean what I want to be there
[12/8/2004 10:03 AM] <Harsha> Bye Srinath
[12/8/2004 10:04 AM] <chathura> and extend it to provide the functionalities required by the engine guys
[12/8/2004 10:04 AM] <alek_s> i will send chat log
[12/8/2004 10:04 AM] <alek_s> bye
[12/8/2004 10:04 AM] |<-- Srinath has left irc.freenode.net ("Client Exiting")
[12/8/2004 10:04 AM] <Deepal> bye all

ChatAgenda/20041208/ChatLog (last edited 2009-09-20 22:47:17 by localhost)