topics discussed: *[Ii]mpl, OmNamespace and QName usage in OM, M1 and testing

-- AleksanderSlominski

Other Chats

[12/22/2004 9:01 AM] <Srinath> Just in time (JIT) :) it is 7.59 here 
[12/22/2004 9:01 AM] -->| Harsha (Harsha@CPE-147-10-36-169.vic.bigpond.net.au) has joined #apache-axis
[12/22/2004 9:01 AM] <Chinthaka> hi all
[12/22/2004 9:01 AM] <alek_s> good that this week meeting is at 9AM EST 
[12/22/2004 9:01 AM] <Jaliya> Hi All
[12/22/2004 9:02 AM] <Harsha> Hi All
[12/22/2004 9:02 AM] <alek_s> i am in europe so it is 3PM but otherwise it would be 4 AM :)
[12/22/2004 9:02 AM] <Ajith> hello all
[12/22/2004 9:02 AM] <alek_s> he all
[12/22/2004 9:02 AM] <gdaniels> good ${TIMEOFDAY}, all
[12/22/2004 9:02 AM] <Deepal> hi
[12/22/2004 9:02 AM] <Srinath> what is 4 today?
[12/22/2004 9:02 AM] <Srinath> shall I start with package structure ?
[12/22/2004 9:02 AM] -->| MHansen (~khookguy@pool-151-205-104-129.ny325.east.verizon.net) has joined #apache-axis
[12/22/2004 9:02 AM] -->| JaimeM (~jmeritt@psc.progress.com) has joined #apache-axis
[12/22/2004 9:03 AM] <gdaniels> Sure.  I'd also like to discuss OMNamespace a little
[12/22/2004 9:03 AM] <Chinthaka> :(
[12/22/2004 9:03 AM] <gdaniels> And I'm sure we can talk engine stuff a lot :)
[12/22/2004 9:03 AM] <Ajith> alek : you back in poland?
[12/22/2004 9:03 AM] <alek_s> yes i am :)
[12/22/2004 9:03 AM] <Ajith> cool
[12/22/2004 9:03 AM] <gdaniels> Eran: why the frowny-face?
[12/22/2004 9:03 AM] <alek_s> so we are spanning minimum 3 continents
[12/22/2004 9:03 AM] <Srinath> shall we have the org.apache.axis.engine only with *Impl for impl classes
[12/22/2004 9:03 AM] <Chinthaka> just kidding :)
[12/22/2004 9:04 AM] <gdaniels> Srinath: +1
[12/22/2004 9:04 AM] <Chinthaka> alek : great isn't it ? :)
[12/22/2004 9:04 AM] <MHansen> Hi - I am new to this chat.  I'm a heavy axis1.2 user and interested in 2.0.  OK if I participate?
[12/22/2004 9:04 AM] <alek_s> yes ;-)
[12/22/2004 9:04 AM] <Ajith> alek : 4 continents (with harsha at australia)
[12/22/2004 9:04 AM] <Srinath> :) what others think?
[12/22/2004 9:04 AM] <alek_s> Hi MHansen!
[12/22/2004 9:04 AM] <gdaniels> MHansen - of course!  Welcome!
[12/22/2004 9:04 AM] <Jaliya> Hi MHansen, welcome
[12/22/2004 9:05 AM] <MHansen> Great.  Thanks :-)
[12/22/2004 9:05 AM] <Deepal>  Hi MHansen
[12/22/2004 9:05 AM] <Ajith> yep , all interested parties are welcome
[12/22/2004 9:05 AM] <Harsha> Hi Mhansen
[12/22/2004 9:05 AM] <Ajith> shall we start then
[12/22/2004 9:05 AM] -->| farhaan (~Miranda@203.143.53.74) has joined #apache-axis
[12/22/2004 9:05 AM] <gdaniels> My compatriot Jaime (who did the JMS stuff for Axis 1.X) is here too, though he's not saying much (which is a little unusual for him, I can tell you :))
[12/22/2004 9:06 AM] <gdaniels> Ajith: We started - Srinath asked about package structure
[12/22/2004 9:06 AM] <gdaniels> :)
[12/22/2004 9:06 AM] <alek_s> MHansen: what are your interests? What is that you want to see in AXIS2 that is not in AXIS1.x ?
[12/22/2004 9:06 AM] <wizard42> Hi * ! I am also is new to this chat. I just would like to listen and have a look where Axis will go to.
[12/22/2004 9:06 AM] <Srinath> shall I sealed the package discussion and change it ASAP ..the silance is assumed as a approval ;)
[12/22/2004 9:07 AM] <gdaniels> Sounds good to me, Srinath.
[12/22/2004 9:07 AM] <Srinath> 3.2.
[12/22/2004 9:07 AM] <Srinath> 1
[12/22/2004 9:07 AM] <Srinath> sealed :)
[12/22/2004 9:08 AM] <Jaliya> ok, the next is the engine, I guess
[12/22/2004 9:08 AM] <alek_s> i think that decision about .imp versus *Impl should be left to particular cases
[12/22/2004 9:08 AM] <MHansen> alek_s: - I'm working on an EAI (enterprise application integration) middleware system that is build around JAX-RPC and other J2EE standards.  My focus with Axis 1.2 has been SAAJ and JAX-RPC.  I'm trying to help get bugs out when I have time.  For Axis-2, I'd like to see JAX-RPC 2.0, a rock-solid SAAJ implementation, and performance improvements.  I'd love to lend a hand as time permits.
[12/22/2004 9:08 AM] <JaimeM> ha, sorry about my quiet nature this morning
[12/22/2004 9:08 AM] <alek_s> if you have more classes than *Impl it iwll b e very messy if you mix interfaces and implementaiton in the same package IMHO
[12/22/2004 9:08 AM] <Srinath> Hi MHansen, JamieM, wizard42 .. welcome all
[12/22/2004 9:09 AM] <alek_s> MHansen: sounds good
[12/22/2004 9:09 AM] <gdaniels> I'm fine with engine.impl.* as well - just don't like axis.impl.engine...
[12/22/2004 9:09 AM] <Srinath> I am +1 for both ..:)
[12/22/2004 9:09 AM] <Jaliya> +1 glen , let's go ahead with engine.impl
[12/22/2004 9:10 AM] <Deepal> yep its better
[12/22/2004 9:10 AM] <alek_s> +1 to both
[12/22/2004 9:10 AM] <gdaniels> discretion of the builder to use package.FooImpl or package.impl.Foo
[12/22/2004 9:11 AM] <Srinath> I see that better not to have impl in package stucture at all?
[12/22/2004 9:11 AM] <Srinath> are we for impl then ?
[12/22/2004 9:11 AM] <gdaniels> Didn't understand that Srinath?
[12/22/2004 9:11 AM] <alek_s> Srinath: if implementation is complex not just *Impl classes where you want ot put additional classes?
[12/22/2004 9:12 AM] <Srinath> Glen: Dr Sanjiva Said +1 from me too. Never liked "impl" in a package name .. just like
[12/22/2004 9:12 AM] <Srinath> I hated "enum" in a package name ;-).
[12/22/2004 9:12 AM] <Srinath> others do nto specify which one 
[12/22/2004 9:12 AM] <Srinath> so I think it is XXImpl
[12/22/2004 9:13 AM] <Srinath> shall we go for o.a.a.engine.impl ect then?
[12/22/2004 9:14 AM] -->| {eng}bar4ka (~marcos@200.216.221.147) has joined #apache-axis
[12/22/2004 9:14 AM] <gdaniels> I'm fine with either way, just not axis.impl.... :)
[12/22/2004 9:14 AM] <Srinath> me too Jaliya, other are we for engine.impl then?
[12/22/2004 9:15 AM] <Jaliya> ??
[12/22/2004 9:15 AM] <gdaniels> okay... let's move on then?
[12/22/2004 9:16 AM] <gdaniels> I'd like to talk about namespaces a little
[12/22/2004 9:16 AM] <gdaniels> I'm a little concerned about this OMNamespace thing
[12/22/2004 9:16 AM] <alek_s> namespaces are always little tricky ...
[12/22/2004 9:16 AM] <{eng}bar4ka> hey guys i am having some problem loading AxisServlet, i guess its not a classpath issue cause HappyAxis.jsp can find it, but i could not load it i guess
[12/22/2004 9:17 AM] <gdaniels> {eng}bar4ka: This chat right now is about Axis2 development, so we're trying to stay on that.  You could either hang out later, or ask on axis-user....
[12/22/2004 9:18 AM] <gdaniels> So here's my concern - there are a lot of cases where I want to use just QNames, and never worry about explicit namespace declaration.
[12/22/2004 9:18 AM] <alek_s> example?
[12/22/2004 9:18 AM] <gdaniels> OMElement foo = new OMElement(qname, content);  writer.write(foo);
[12/22/2004 9:18 AM] <alek_s> content?
[12/22/2004 9:19 AM] <gdaniels> should gen something like <ns:qname xmlns:ns="ns">content</ns:qname>
[12/22/2004 9:19 AM] <alek_s> you mean element text content or List<OmElement>
[12/22/2004 9:19 AM] <Chinthaka> seems like some Sri lAnkan guys have some prob
[12/22/2004 9:19 AM] <Chinthaka> with chat
[12/22/2004 9:19 AM] <gdaniels> but that's secondary
[12/22/2004 9:19 AM] <Chinthaka> Srinath, Deepal and Ajith
[12/22/2004 9:19 AM] <gdaniels> the point is I DON'T want to have to declare an OMNamepace
[12/22/2004 9:19 AM] <gdaniels> Eran: oh dear
[12/22/2004 9:20 AM] <Chinthaka> they just told me, but lets go ahead
[12/22/2004 9:20 AM] <gdaniels> aren't they using the same net connection as you?
[12/22/2004 9:20 AM] <alek_s> if you depend on namespace prefix you better have prefix in QName right?
[12/22/2004 9:20 AM] <Jaliya> we are in two location glen
[12/22/2004 9:20 AM] <Chinthaka> Glen : no
[12/22/2004 9:21 AM] <gdaniels> Alek: if you depend on a particular prefix, yes
[12/22/2004 9:21 AM] -->| Srinath_ (~Srinath@220.247.247.133) has joined #apache-axis
[12/22/2004 9:21 AM] -->| deepal123 (~deepal@220.247.247.133) has joined #apache-axis
[12/22/2004 9:21 AM] <gdaniels> but you can do that : QName qname = new QName("ns", "foo", "prefix")
[12/22/2004 9:21 AM] <gdaniels> still no need to make OMNamespace
[12/22/2004 9:21 AM] <deepal123> we had some prob.
[12/22/2004 9:21 AM] -->| Ajith1030 (~Miranda@220.247.247.133) has joined #apache-axis
[12/22/2004 9:21 AM] <gdaniels> I just want to make sure the APIs work without it
[12/22/2004 9:21 AM] <Chinthaka> but glen 
[12/22/2004 9:22 AM] <Chinthaka> isn't it better to have OMNamespace
[12/22/2004 9:22 AM] <Ajith1030> aah finally Back in the chat again
[12/22/2004 9:22 AM] <deepal123> what did u talk , can some put that into yahoo
[12/22/2004 9:22 AM] <Chinthaka> so that other Elements, attr, etc can use that
[12/22/2004 9:22 AM] <gdaniels> Eran: Clearly, in my opinion the answer is "no". :)  Or at least "not always".
[12/22/2004 9:22 AM] <Jaliya> Deepal: Sent the chat using yahoo
[12/22/2004 9:23 AM] <gdaniels> If you care about it, sure you should be able to register particular NS mappings at particular places.  I definitely want that.
[12/22/2004 9:23 AM] <gdaniels> I just don't want to force people to use this class.
[12/22/2004 9:23 AM] <alek_s> Glen: i thin in this case it is as if you delcared namespace on this element (shows up during serialization) 
[12/22/2004 9:23 AM] <gdaniels> Alek: Yes.
[12/22/2004 9:23 AM] <gdaniels> Another example:
[12/22/2004 9:24 AM] |<-- Harsha has left irc.freenode.net ()
[12/22/2004 9:24 AM] <gdaniels> QName q1 = new QName("ns1", "foo")
[12/22/2004 9:24 AM] <alek_s> Glen: the only a problem is if you try later to declare the same prefix to bind ieth other namespace - but that is an error that OM impl should detect
[12/22/2004 9:24 AM] <gdaniels> QName q1 = new QName("ns1", "bar")
[12/22/2004 9:24 AM] |<-- Deepal has left irc.freenode.net (Read error: 60 (Operation timed out))
[12/22/2004 9:24 AM] |<-- Srinath has left irc.freenode.net (Read error: 60 (Operation timed out))
[12/22/2004 9:24 AM] <gdaniels> OMElement root = new OMElement(q1);
[12/22/2004 9:24 AM] <gdaniels> OMElement child = new OMElement(q2);
[12/22/2004 9:24 AM] <gdaniels> root.addChild(child);
[12/22/2004 9:24 AM] <gdaniels> should serialize to
[12/22/2004 9:24 AM] <alek_s> Glen: in this case there is *no* namespace prefix declared
[12/22/2004 9:25 AM] <gdaniels> <p1:foo xmlns:p1="ns1"><p1:bar/></p1:foo>
[12/22/2004 9:25 AM] <Ajith1030> Hey What did you guys decide about the OMNamespace
[12/22/2004 9:25 AM] <gdaniels> yes
[12/22/2004 9:25 AM] <alek_s> StAX or XmlPull is able to generate you prefix but it is not known until serialization happen
[12/22/2004 9:25 AM] <alek_s> Glen: exactly
[12/22/2004 9:25 AM] <gdaniels> so the framework should declare you one
[12/22/2004 9:25 AM] <alek_s> however if you put root into OMElement root2=new OMElement(otherQname)
[12/22/2004 9:26 AM] <alek_s> and otherQName has prefix "n" then you get serializaiton where <n:foo><n:bar/></n:foo> 
[12/22/2004 9:26 AM] <alek_s> is it what you want?
[12/22/2004 9:26 AM] <gdaniels> there are tradeoffs
[12/22/2004 9:27 AM] <gdaniels> If I'm building up content, and I want to write a QName string inside an attribute or something, it's nice to have the mapping done already
[12/22/2004 9:27 AM] <alek_s> i think it is the simplest way to do fragments of XML - elements that are not attached to XML document
[12/22/2004 9:27 AM] <alek_s> then you *really* should declare mapping explicitly
[12/22/2004 9:27 AM] <{eng}bar4ka> gdaniels, sorry about "offtopic", but the axis-user you talk is a channel or the mailing list ?
[12/22/2004 9:27 AM] <gdaniels> We don't now and it works well
[12/22/2004 9:28 AM] <gdaniels> {eng}bar4ka: Mailing list
[12/22/2004 9:28 AM] <alek_s> otherwise effect iis unpredictable
[12/22/2004 9:28 AM] <gdaniels> maybe we can have an option
[12/22/2004 9:28 AM] <alek_s> if you do not declare namespace ...
[12/22/2004 9:28 AM] <alek_s> for attributes it would  be neat to have option to set QName as attribute value
[12/22/2004 9:29 AM] <Ajith1030> Hey guys, shall we get this namespace thing into the mailing list
[12/22/2004 9:29 AM] <alek_s> and let OM to resolve prefix during serialization
[12/22/2004 9:29 AM] <gdaniels> alek: Axis 1.X does that
[12/22/2004 9:29 AM] <alek_s> (i have done this in XSUL/XPP3/XB1 and it works pretty well)
[12/22/2004 9:29 AM] <gdaniels> well something like it
[12/22/2004 9:29 AM] <gdaniels> I'm +1 for it
[12/22/2004 9:29 AM] <Ajith1030> it seems we need to dicuss a bit about the M1 
[12/22/2004 9:30 AM] <alek_s> Glen: in AXIS 1.x you talk about serializer writing to SAX?
[12/22/2004 9:30 AM] <alek_s> Glen: or DOM?
[12/22/2004 9:30 AM] <gdaniels> Alek: In Axis1.1 it writes to text
[12/22/2004 9:30 AM] <gdaniels> s/1.1/1.X/
[12/22/2004 9:31 AM] <{eng}bar4ka> gdaniels, sorry about bothering again, i will hang for a hope and help to solve this problem, its already in the axis-user mailing list at http://marc.theaimsgroup.com/?l=axis-user&m=110296693327208&w=2
[12/22/2004 9:31 AM] <alek_s> Glen: what part of AXIS 1.x do you refer to exactly?
[12/22/2004 9:31 AM] <gdaniels> SerializationContext (but really also MessageElement)
[12/22/2004 9:32 AM] <gdaniels> MessageElement has a list of "qname-attributes" which need to be resolved at serialization time
[12/22/2004 9:32 AM] <alek_s> so it is geared for SAX writer we are talking about tight?
[12/22/2004 9:33 AM] <gdaniels> ?
[12/22/2004 9:33 AM] <alek_s> you get exactly the same gfunctionality if you declare set of namespaces on top-level elemenbt in OM (liek SOAP:Envelope)
[12/22/2004 9:33 AM] <alek_s> Geln: when qname-attributes are resolved?
[12/22/2004 9:34 AM] <alek_s> I am fine to move discussion to mailing list
[12/22/2004 9:34 AM] <gdaniels> I don't remember exactly, Alek
[12/22/2004 9:34 AM] <gdaniels> Need to go look
[12/22/2004 9:34 AM] <Chinthaka> Alek : I also agree :)
[12/22/2004 9:34 AM] <gdaniels> The point is that the idea of attributes with qname values is good
[12/22/2004 9:34 AM] <alek_s> lets then talk about M1 ?
[12/22/2004 9:34 AM] |<-- Ajith has left irc.freenode.net (Read error: 113 (No route to host))
[12/22/2004 9:35 AM] <alek_s> Glen: i have seen that used in XSUL and i a gree it is GooThing(tm) :)
[12/22/2004 9:35 AM] <Chinthaka> Ajith quit again ???? :( :(
[12/22/2004 9:35 AM] <gdaniels> Ajith1030 is Ajith :)
[12/22/2004 9:35 AM] <gdaniels> That was Ajith's ghost departing
[12/22/2004 9:35 AM] <Chinthaka> :)
[12/22/2004 9:36 AM] <gdaniels> So Ajith, what did you want to talk about re: M1?
[12/22/2004 9:37 AM] <gdaniels> Hm
[12/22/2004 9:37 AM] <gdaniels> maybe he did lose the connection again
[12/22/2004 9:37 AM] <Chinthaka> Ajith, Srinath, Farhaan, u there ?
[12/22/2004 9:38 AM] <farhaan> I am here, but Srinath & Co. seems to be having some problems at the University
[12/22/2004 9:38 AM] <Srinath_> we are here :)
[12/22/2004 9:38 AM] <gdaniels> darn Internet.
[12/22/2004 9:38 AM] <Chinthaka> yeah Glen
[12/22/2004 9:38 AM] <Ajith1030> hold on a sec till I align my name
[12/22/2004 9:38 AM] <Chinthaka> this always happens in Wednesday nights after 8
[12/22/2004 9:39 AM] <Chinthaka> :(
[12/22/2004 9:39 AM] <gdaniels> Oh, btw - I won't be on next week's chat as I'll be in London
[12/22/2004 9:39 AM] <Srinath_> oops 
[12/22/2004 9:40 AM] <Chinthaka> "Glen in London", "Alek in Poland", "We in Srilanka, always" :D
[12/22/2004 9:40 AM] <Srinath_> we can talk bit M1 ..
[12/22/2004 9:41 AM] <Srinath_> shall we move the OM code to src ?
[12/22/2004 9:41 AM] <gdaniels> Let's do the move to src after M1
[12/22/2004 9:41 AM] =-= Ajith1030 is now known as Ajith
[12/22/2004 9:41 AM] <gdaniels> ?
[12/22/2004 9:41 AM] <alek_s> Glen: would it make any difference if next week chat was 9AM (ie. 3PM in England)?
[12/22/2004 9:41 AM] <gdaniels> build off prototype2 for now
[12/22/2004 9:42 AM] <gdaniels> alek: Yes, that would work better
[12/22/2004 9:42 AM] <Ajith> that will save us a whole lot of trouble :)
[12/22/2004 9:42 AM] <alek_s> what is M1 exactly?
[12/22/2004 9:42 AM] <alek_s> and what is "src" in this context?
[12/22/2004 9:42 AM] <gdaniels> alek: src is the "real" src directory
[12/22/2004 9:42 AM] <Srinath_> yes let's do it in p2 then
[12/22/2004 9:42 AM] <Chinthaka> I think Farhaan sent an email
[12/22/2004 9:42 AM] <gdaniels> i.e. not scratch/* but java/src
[12/22/2004 9:42 AM] <Srinath_> alek ..src is the real tree
[12/22/2004 9:42 AM] <Chinthaka> to mailing list abt M1 :(
[12/22/2004 9:43 AM] <Srinath_> M1 = Milstone 1 I think ;)
[12/22/2004 9:43 AM] <Srinath_> I think we can start revirwing the code now 
[12/22/2004 9:44 AM] <Srinath_> for everyhing
[12/22/2004 9:44 AM] <gdaniels> reviewing?
[12/22/2004 9:44 AM] <alek_s> did he send it to [Axis2]? i can not find this ... anyway i will find it
[12/22/2004 9:44 AM] <Srinath_> review :)
[12/22/2004 9:45 AM] <Ajith> ahum a review just now
[12/22/2004 9:45 AM] <Ajith> ?
[12/22/2004 9:45 AM] <Chinthaka> Alek : http://wiki.apache.org/ws/FrontPage/Axis2/Release01M1
[12/22/2004 9:45 AM] <Srinath_> I put most of the things we argeed to p2 (expect movinf address logic to sender)
[12/22/2004 9:45 AM] <Ajith> I am doing some major changes to the code so I need at least a day to clean the things up
[12/22/2004 9:45 AM] <Chinthaka> [Axis2] Milestone Release & To Do's 
[12/22/2004 9:45 AM] <Srinath_> but there may be missing things
[12/22/2004 9:46 AM] <alek_s> Chinthaka: thanks!
[12/22/2004 9:46 AM] <Srinath_> e.g. ClientAPI added by Jaliya and Co last week :)
[12/22/2004 9:47 AM] <Jaliya> Deepal and I have added both the sync and async paths for the client
[12/22/2004 9:47 AM] <gdaniels> I think everyone should try to take a serious look at the code as it stands this week/weekend
[12/22/2004 9:47 AM] <Jaliya> not the ideal async with a seperate listener but using callbacks
[12/22/2004 9:47 AM] <Ajith> agreed for this weekend :)
[12/22/2004 9:47 AM] <gdaniels> Jaliya: +1
[12/22/2004 9:47 AM] <Srinath_> +1 glen .. I try my best to clean up testCases ect by this week
[12/22/2004 9:48 AM] <gdaniels> Srinath_: Did you build an infrastructure for doing client-server testing?
[12/22/2004 9:48 AM] <farhaan> Alek:section 4 on Release01M1 needs updating though 
[12/22/2004 9:48 AM] <gdaniels> i.e. using ant's <parallel> task
[12/22/2004 9:48 AM] <Srinath_> we all try to optimise thing till friday and try to freeze the code
[12/22/2004 9:48 AM] <deepal123> im sorry im bit busy with deployment stuff
[12/22/2004 9:48 AM] <Jaliya> we need to clarify that I guess, the testing of scenarios
[12/22/2004 9:49 AM] |<-- JaimeM has left irc.freenode.net ()
[12/22/2004 9:49 AM] <Srinath_> glen: I did it using simple axis server started at the setup() for now
[12/22/2004 9:49 AM] <Jaliya> Can we do it using ant or just using simple axis server
[12/22/2004 9:49 AM] <gdaniels> Jaliya: Um, both. :)
[12/22/2004 9:49 AM] <Jaliya> we are doing the same glen at the moment
[12/22/2004 9:49 AM] <Srinath_> I think the <paralel> info is at the 1.x test .. will have a look
[12/22/2004 9:49 AM] <gdaniels> 1.X doesn't use <parallel>
[12/22/2004 9:49 AM] <gdaniels> although it could
[12/22/2004 9:50 AM] <gdaniels> right now we use a custom task I think
[12/22/2004 9:50 AM] <alek_s> Farhaan: how do you plan to do testing?
[12/22/2004 9:51 AM] <Srinath_> I usually like to start alll in the setup of the TestCxase including simple AxisServer!
[12/22/2004 9:51 AM] <alek_s> Farhaan: what features are to be tested? 
[12/22/2004 9:51 AM] <Chinthaka> gle
[12/22/2004 9:51 AM] <Srinath_> glen:what does parellel do?g
[12/22/2004 9:52 AM] <farhaan> Alek: we have already come up with unit tests and most probably we want to include some of the whitemesa tests next week
[12/22/2004 9:52 AM] <alek_s> Farhaan:i think it would be useful to add this to wiki page
[12/22/2004 9:52 AM] <gdaniels> The parallel task just runs things on multiple threads and syncs them
[12/22/2004 9:53 AM] <alek_s> (in 3.6 Test Cases)
[12/22/2004 9:53 AM] <farhaan> Alek: agreed, i will add the test matrix
[12/22/2004 9:53 AM] <gdaniels> Here's what I'd like to see - there are a variety of client tests we're going to have, and they might want to run against a variety of servers (in-proc SimpleAxisServer, out-of-proc SimpleAxisServer, servlet engine, etc)
[12/22/2004 9:54 AM] <Jaliya> glen : what is in-proc
[12/22/2004 9:54 AM] <Srinath_> glen:you mean to set up a one or more servers and send all requests to them
[12/22/2004 9:54 AM] <gdaniels> So there should be a generic "setup" which keys off config/switches that starts up anything it needs to (SimpleAxisServer), and then the client tests receive a base URL from that configuration which points them at the right place (either "local:" or "http:"...)
[12/22/2004 9:54 AM] <gdaniels> Srinath_: I mean one server, but which one is dependent on config.
[12/22/2004 9:55 AM] <Srinath_> glen:I got you
[12/22/2004 9:55 AM] <gdaniels> typical cases are 1) servlet engine, so setup is external and we just need URL; 2) SimpleAxisServer needs to run with HTTP; 3) use local in-process server with "local:" transport
[12/22/2004 9:55 AM] <alek_s> if you have SimpleAxisServer using JUNIT.setUp/teardown to start/stop server is easy
[12/22/2004 9:56 AM] <gdaniels> alek: but we don't want to start/stop SAS for each test
[12/22/2004 9:56 AM] <Srinath_> we can have both types ..in the junit test cases we can use setup(..) to bring up the servers
[12/22/2004 9:56 AM] <gdaniels> We want to start one at the beginning of the run and then stop it after all tests
[12/22/2004 9:56 AM] <Srinath_> but we can test samples using one server
[12/22/2004 9:57 AM] <Chinthaka> Glen, one thing, what is in-proc and out-of-proc ? 
[12/22/2004 9:57 AM] <gdaniels> in-process, out-of-process
[12/22/2004 9:57 AM] <alek_s> GLen: why not - those are unit tests
[12/22/2004 9:57 AM] <gdaniels> Because it'll be sloooooow
[12/22/2004 9:57 AM] <deepal123> glen with Async one we have some problems for testing
[12/22/2004 9:57 AM] <alek_s> Glen: i do it all the time in XSUL - if you have really small server starting/stopping it is very fast ...
[12/22/2004 9:57 AM] <Srinath_> glen:alek: I thik we should have both types
[12/22/2004 9:57 AM] <alek_s> Glen: how slow is sloooooow :)
[12/22/2004 9:58 AM] <gdaniels> alek: yes, but as the # of tests increases even a small incremental slowdown can snowball
[12/22/2004 9:58 AM] <deepal123> that is in async case we start a new thread and assign work to that thread
[12/22/2004 9:58 AM] <alek_s> i start server, bomabrd it with methods, shutdown when it is finished
[12/22/2004 9:58 AM] <Srinath_> plus I have seen Geronimo brings up complete j2ee contianer in unit test setup() :)
[12/22/2004 9:58 AM] <alek_s> i have pretty complicated async testing done like that
[12/22/2004 9:58 AM] <Jaliya> + 1 for Glen one server for some tests
[12/22/2004 9:59 AM] <gdaniels> I need to run - got a meeting..!
[12/22/2004 9:59 AM] <gdaniels> Talk more on mailing list
[12/22/2004 9:59 AM] <Chinthaka> oki
[12/22/2004 9:59 AM] <gdaniels> bye all
[12/22/2004 9:59 AM] <Srinath_> sure glen .. will put a mail regarding test 
[12/22/2004 9:59 AM] <Jaliya> but I think we have to write the test cases with some connections
[12/22/2004 9:59 AM] <Chinthaka> but these days mailing list seems silent :)
[12/22/2004 9:59 AM] |<-- gdaniels has left irc.freenode.net ()
[12/22/2004 10:00 AM] <alek_s> Srinath: i think it is what setup() is supposed to do in *unit* test...
[12/22/2004 10:00 AM] <Chinthaka> I also got to go, byee all
[12/22/2004 10:00 AM] -->| dims (~dims@exchange.sun.com) has joined #apache-axis
[12/22/2004 10:00 AM] <Chinthaka> today log ?
[12/22/2004 10:00 AM] <alek_s> i have even bringing up two servers in unti tests and it worked just fine (http://www.extreme.indiana.edu/viewcvs/~checkout~/xsul/java/tests/xsul/msg_box/TestMsgBox.java)
[12/22/2004 10:00 AM] <alek_s> i will send it 
[12/22/2004 10:01 AM] <alek_s> you did it last week
[12/22/2004 10:01 AM] =-= Chinthaka is now known as Chinthaka_away
[12/22/2004 10:01 AM] <--| deepal123 has left #apache-axis

ChatAgenda/20041222/ChatLog (last edited 2009-09-20 22:48:27 by localhost)