You are viewing a plain text version of this content. The canonical link for it is here.
Posted to java-dev@axis.apache.org by Aleksander Slominski <as...@cs.indiana.edu> on 2004/12/02 05:32:12 UTC
[Axis2] IRC chat log 2004-12-01
today we discussed how to make AXIS2 smarter and decided to put
more AIs into AXIS2 ... Action Items:
1. Farhaan: outline of Release plan for O1 in wiki (revised
http://wiki.apache.org/ws/FrontPage/Axis2/ProjectPlan ?)
we discussed what needs/can be released and currently looks like OM
(after it is not SOAP specific) and set of samples/test cases for:
interop client/server scenarios (http://www.whitemesa.net/),
data binding examples (Castor, XmlBeans, ...) - how to do
flexible data binding over OM that has easy to use API,
and xml parsing of xmlschema tests collection
(http://www.w3.org/2001/05/xmlschema-test-collection/)?
Also very weak agreement that modularity is good ...
next chat:
December 08, 2004 9am EST *on* WEDNESDAY (That's 8pm Wed in SriLanka)
update agenda for next week in
http://wiki.apache.org/ws/ChatAgenda
thanks,
alek
[12/1/2004 10:01 PM] <gdaniels> Good evening/morning, all
[12/1/2004 10:02 PM] <Deepal> hi GM/GE
[12/1/2004 10:02 PM] -->| Ajith (~Miranda@220.247.249.192) has joined #apache-axis
[12/1/2004 10:02 PM] <Ajith> hi all
[12/1/2004 10:03 PM] <Srinath> good morning all :)
[12/1/2004 10:03 PM] <gdaniels> I updated the script to indicate that the chat is the NEXT morning in
Sri Lanka the next time it comes around, btw.
[12/1/2004 10:03 PM] <alek_s> GM/E to everybody (GME as another TLA?)
[12/1/2004 10:03 PM] <dims> hi all
[12/1/2004 10:04 PM] <Srinath> what are in for today?
[12/1/2004 10:04 PM] <gdaniels> good Q
[12/1/2004 10:05 PM] <gdaniels> probably should talk about general scheduling / release stuff
[12/1/2004 10:05 PM] <Srinath> yap :)
[12/1/2004 10:05 PM] <alek_s> it seems nobody bother to put any agenda so maybe we call off it this
meeting - meetings without agendas are waste fo time ...
[12/1/2004 10:05 PM] <alek_s> http://wiki.apache.org/ws/ChatAgenda/20041201
[12/1/2004 10:06 PM] <alek_s> Sanjiva proposed to do release by the end of the year? sounds like
another topic to discuss?
[12/1/2004 10:06 PM] <Srinath> when we start releasing and sheduling it will cover all the time ;)
[12/1/2004 10:06 PM] <gdaniels> alek : disagree pretty strongly. Agendas are good but we have way too
much on our plate at this point to worry about extended silences. :)
[12/1/2004 10:06 PM] <Srinath> yap :)
[12/1/2004 10:07 PM] <alek_s> i wish we were using mailing list more to hash issues
[12/1/2004 10:07 PM] <gdaniels> Agreed - although I must admit I actually really like the immediacy of
IRC type things
[12/1/2004 10:07 PM] -->| Chinthaka (~EC@220.247.240.170) has joined #apache-axis
[12/1/2004 10:07 PM] <Chinthaka> hi all
[12/1/2004 10:08 PM] <gdaniels> I find it much easier to work through many kinds of problems in real
time than via email
[12/1/2004 10:08 PM] -->| Jaliya (~Miranda@220.247.240.170) has joined #apache-axis
[12/1/2004 10:08 PM] <Srinath> yap glen; we all like if answers come quickly
[12/1/2004 10:08 PM] <gdaniels> tradeoffs
[12/1/2004 10:08 PM] <gdaniels> Where is that Weerawarana guy anyway
[12/1/2004 10:08 PM] <Chinthaka> :)
[12/1/2004 10:08 PM] <Jaliya> :)
[12/1/2004 10:08 PM] <Srinath> where we start at releasing?
[12/1/2004 10:08 PM] <alek_s> another meeting/conference/call/...?
[12/1/2004 10:08 PM] <Chinthaka> thats good
[12/1/2004 10:09 PM] <gdaniels> Well if Sanjiva doesn't show up, we don't have to listen to his
proposal and we can get back to real work without worrying about releasing yet. :)
[12/1/2004 10:09 PM] <Ajith> :)
[12/1/2004 10:09 PM] <alek_s> :-)
[12/1/2004 10:09 PM] <Deepal> :)
[12/1/2004 10:09 PM] <gdaniels> So here's what I think, to start things off - I don't think honestly
that we're ready for a release of the "whole potato" yet.
[12/1/2004 10:10 PM] <gdaniels> I think there are too many murky/unclear parts of the design to put it
out there
[12/1/2004 10:10 PM] <alek_s> personally i would liek things to be modular/componentized so it is
easier to work on pieces
[12/1/2004 10:10 PM] <gdaniels> +1 Alek
[12/1/2004 10:10 PM] <Ajith> Glen : Sanjiva had a good reason. He said dozen other guys are doing SOAP
stacks and if we dont put out something soon we are going to loose a bigcommunity
[12/1/2004 10:10 PM] <gdaniels> I'd love to see us do a near-term release of just OM (incl. hooks for
MTOM/DataBinding)
[12/1/2004 10:11 PM] <Srinath> I think we should take a reduced core .. go though the stuff and release
[12/1/2004 10:11 PM] <Chinthaka> but glen, this can be completely thrown away, if something is wrong
[12/1/2004 10:11 PM] |<-- ChanServ has left irc.freenode.net (ACK! SIGSEGV!)
[12/1/2004 10:11 PM] <alek_s> however i am in real time crunch for next two weeks so i am not sure if
i can get that much done
[12/1/2004 10:11 PM] <gdaniels> Ajith: I understand that, but I'm not sure I agree with it - or rather
I'm unsure that a few more weeks of tightening won't make things a lot more comfortable
[12/1/2004 10:11 PM] <Srinath> limited but enough to keep the user comminuity on there toes:)
[12/1/2004 10:11 PM] <alek_s> we should not raise too quickly peopel expectations of AXIS2
[12/1/2004 10:11 PM] <Ajith> Srinath : I guess that is what Sanjiva wanted!
[12/1/2004 10:11 PM] <alek_s> it should be understood ti is early prototype so it may not work etc
[12/1/2004 10:12 PM] <gdaniels> alek: Yes, but we should also demonstrate that we're getting good stuff
thought about.
[12/1/2004 10:12 PM] <gdaniels> I think showing people our designs / architecture ideas is as important
as releasing code at this stage
[12/1/2004 10:12 PM] -->| ChanServ (ChanServ@services.) has joined #apache-axis
[12/1/2004 10:12 PM] <gdaniels> That can also encourage people to hop on board
[12/1/2004 10:13 PM] <alek_s> i agree that is positive aspect and may be well worht trying to do this
in this context ...
[12/1/2004 10:13 PM] <gdaniels> That said, I'm not AGAINST a release, I just don't think we're ready
yet. Agree with Srinath about reduced core, though - if we could trim it to the parts we can agree on....
[12/1/2004 10:13 PM] <Deepal> My idea is to get better undersatdning about axis 2 , it is good if we
can do that
[12/1/2004 10:13 PM] <gdaniels> Deepal: When you get that understanding, please let the rest of us know
how it works, OK? :)
[12/1/2004 10:14 PM] <Deepal> :)
[12/1/2004 10:14 PM] <gdaniels> So what do other people think about Alek's modularity point?
[12/1/2004 10:14 PM] <gdaniels> Does it make sense to release OM as a separately buildable
mini-package? (regardless of what else we release)
[12/1/2004 10:14 PM] <gdaniels> I think +1
[12/1/2004 10:14 PM] <alek_s> Is anybody interested in making something like AXIS2 XT (extrem cutting
edge technology) and fuse in JDK5, util.concurent APIs, generics, metadata annotations and other cool
stuff and see how easy can be WS programming?
[12/1/2004 10:15 PM] <Jaliya> Is it usable separately, I mean the OM
[12/1/2004 10:15 PM] <Chinthaka> I think OM is in good shape and if we can get engine also to place,
like Srinath said
[12/1/2004 10:15 PM] <alek_s> i would love to see OM as a module
[12/1/2004 10:15 PM] <Srinath> I think wht we ant to put out is a M1 or snapshot
[12/1/2004 10:15 PM] <Ajith> Glen : Modularity is good. Om is already a seperately buildable package :)
[12/1/2004 10:15 PM] <gdaniels> alek: I think we build the core first, and then build ease-of-use stuff
like that on top. I'd like to see it though. :)
[12/1/2004 10:15 PM] <Jaliya> Yes, Engine and OM will go together
[12/1/2004 10:16 PM] <dims> alek: +1 to what glen said.
[12/1/2004 10:16 PM] <gdaniels> Well, here's a suggestion - we could do a first release of just OM.
[12/1/2004 10:16 PM] <gdaniels> Get that out, then the next step is to add minimal engine stuff around it.
[12/1/2004 10:16 PM] <Chinthaka> just OM ??
[12/1/2004 10:16 PM] <Ajith> hmmm , Would that be ok? I mean is it enough?
[12/1/2004 10:16 PM] <alek_s> Glen: i think that numebr of layers or pretty covering will not make
things easier but may be leaking complexity ...
[12/1/2004 10:17 PM] <alek_s> OM to power Engine
[12/1/2004 10:17 PM] <alek_s> :)
[12/1/2004 10:17 PM] <Srinath> glen what do you think about making in Om + minimal engine to meake user
see it as a soapengine :)
[12/1/2004 10:17 PM] <alek_s> we coudl just use httpclient etc to code some simple samples
[12/1/2004 10:17 PM] <alek_s> so it does actually soemthing
[12/1/2004 10:17 PM] <gdaniels> Just OM. That will let people (lots more people than just us) play
with the API and integrate it into their non-Axis projects if they want
[12/1/2004 10:17 PM] <Chinthaka> I also agree with Srinath;s suggestion
[12/1/2004 10:18 PM] <Srinath> let keep the engine minimal .. remove unclear parts
[12/1/2004 10:18 PM] <gdaniels> Srinath: I agree with you, I just think these points can be separated.
[12/1/2004 10:18 PM] <Ajith> Glen : I think the word "release" is bit too much here. Cant we do a
"release candidate" or "protoype release"
[12/1/2004 10:18 PM] <Ajith> Because I strongly feel Sanjiva has a point
[12/1/2004 10:18 PM] <Srinath> but let let a flow messag erun throuhg handlers
[12/1/2004 10:18 PM] <dims> We can stagger these
[12/1/2004 10:18 PM] <dims> 1: OM
[12/1/2004 10:18 PM] <dims> 2: OM+Engine
[12/1/2004 10:18 PM] <dims> (minimal)
[12/1/2004 10:19 PM] <gdaniels> In other words, why not publish just OM first, get our
release-management abilities working well, and then add a minimal engine soon thereafter
[12/1/2004 10:19 PM] <gdaniels> yes
[12/1/2004 10:19 PM] <Srinath> yes glen , when we call it m1, we should have some knd of soap engine to
user to see
[12/1/2004 10:19 PM] <gdaniels> +1 to dims, in other words
[12/1/2004 10:19 PM] <gdaniels> Srinath: So let's call it O1, then M1 :)
[12/1/2004 10:19 PM] -->| chathura (~chathura@220.247.240.170) has joined #apache-axis
[12/1/2004 10:19 PM] <dims> they can use it just like SAAJ
[12/1/2004 10:19 PM] <Srinath> glen ,yes that resonable
[12/1/2004 10:20 PM] <gdaniels> I still think there's some OM work to do, right? Then there is buiding
tests, getting the maven build stuff cleaned up, adding samples...
[12/1/2004 10:20 PM] <gdaniels> that's a couple of weeks work right there
[12/1/2004 10:20 PM] <gdaniels> In the meanwhile we keep honing our shared understanding of the engine,
and decide exactly what "minimal" means for M1
[12/1/2004 10:20 PM] <Srinath> we have some stuff in prototype 2 .. with some tst case
[12/1/2004 10:20 PM] <Ajith> BTW i guess deepal and Srinath has done a lot with deployment and engine.
[12/1/2004 10:20 PM] <Ajith> so we can include that with little effort
[12/1/2004 10:21 PM] <gdaniels> Ajith: The problem with that is that we don't all agree on it yet, I think.
[12/1/2004 10:21 PM] -->| farhaan (~Miranda@220.247.249.192) has joined #apache-axis
[12/1/2004 10:21 PM] <Srinath> all;then let us start with putting the om in to the src
[12/1/2004 10:21 PM] <Ajith> hmmmm..
[12/1/2004 10:21 PM] <Srinath> and get it tested and shaped up
[12/1/2004 10:22 PM] <Ajith> Glen : the thing is OM is almost done (with some test cases) and so far
seems to work fine
[12/1/2004 10:22 PM] <Srinath> bun not call too much attention for user community just to OM .. they
wll not understand
[12/1/2004 10:22 PM] <gdaniels> +1 to that. Alek, are you happy with the current OM state-of-the-art
enough to +1 checking it into axis2/java/src?
[12/1/2004 10:22 PM] <alek_s> O1 with MTOM targeted for next Ox
[12/1/2004 10:22 PM] <dims> Let's say we use OM one the client side. we should be able to send make a
soap message using OM and send it using java.net.URL*
[12/1/2004 10:22 PM] <gdaniels> dims: +1
[12/1/2004 10:22 PM] <alek_s> i would liek to review OM API to make it work well with MTOM and data
bindings - bit no hurry
[12/1/2004 10:22 PM] <dims> We can pick tests on Whitemesa interop site - http://www.whitemesa.net/
(for the server-side)
[12/1/2004 10:23 PM] <Srinath> dims +1 that is a cool
[12/1/2004 10:23 PM] <gdaniels> Likewise having a simple servlet pull in a message from the servlet's
InputStream into OM
[12/1/2004 10:23 PM] <dims> EXACTLY :)
[12/1/2004 10:23 PM] <Chinthaka> alek : data binding in to OM ?
[12/1/2004 10:23 PM] <gdaniels> Chinthaka: Yes... just like we discussed last week! :)
[12/1/2004 10:23 PM] <Srinath> alek; I think we can differ mtom and databinding
[12/1/2004 10:24 PM] <Srinath> we can do xml in xml out m1
[12/1/2004 10:24 PM] <alek_s> no just to check that data bindnig tools liek Castor, XmlBeans, etc can
easily be integrated to consume and produce OM
[12/1/2004 10:24 PM] <dims> pick the same interop tests and make a servlet based thingy
[12/1/2004 10:24 PM] <gdaniels> "hello world" example is just constructing an OM and serializing it
[12/1/2004 10:24 PM] <gdaniels> next is "echo"
[12/1/2004 10:24 PM] <alek_s> i really liek idea of very small interop service / client impl
[12/1/2004 10:25 PM] <Chinthaka> alek : OM can still be integrated to Castor and other data binding tools
[12/1/2004 10:25 PM] <gdaniels> then a simple transform (watch for a particular elemetn and replace its
content)
[12/1/2004 10:25 PM] <gdaniels> etc
[12/1/2004 10:25 PM] <alek_s> i did integration with XML-Java data binding APIs in past - it is at
minimum tricky and sometimes ugly
[12/1/2004 10:25 PM] <Srinath> Dims routine Q for test cases where can I finds the interop testcase? so
we make them out tetcases
[12/1/2004 10:25 PM] <dims> http://www.whitemesa.net/
[12/1/2004 10:26 PM] <gdaniels> Srinath/dims: I don't see why we should use those test cases for just
OM work... it seems like the wrong level of abstraction.
[12/1/2004 10:26 PM] <Srinath> ppl we are going in to OM :( without finishing relase story
[12/1/2004 10:26 PM] <alek_s> we would target first round 2 base right?
[12/1/2004 10:26 PM] <gdaniels> We won't be doing the SOAP processing model, so it won't "really" be
SOAP at all
[12/1/2004 10:26 PM] <dims> Reason for approach: we need to check if we are FAST enough compared to
other toolkits for the same interop scenarios.
[12/1/2004 10:26 PM] <dims> Which is the reason for doing OM in the first place :)
[12/1/2004 10:26 PM] <gdaniels> dims: but we can't really test that until we have the SOAP engine built.
[12/1/2004 10:27 PM] <gdaniels> I agree speed is important, but XML/HTTP is good enough to test that
[12/1/2004 10:27 PM] <alek_s> simple XML messaging and simple SOAP procseeing can be done quickly
(check mustUnderstand and fault generation - anything more than that needed?!)
[12/1/2004 10:27 PM] <alek_s> no need for full engine for that ?!
[12/1/2004 10:27 PM] <gdaniels> alek: no, not really
[12/1/2004 10:27 PM] <Srinath> yes glen but we need soap message to test OM anyway .. why not just use
introp ones
[12/1/2004 10:28 PM] <alek_s> Glen: what would be missing?
[12/1/2004 10:28 PM] <dims> yes srinath
[12/1/2004 10:28 PM] <farhaan> In that case can we not figure out what is the absolute minimum that we
need in the engine? and target that for M1
[12/1/2004 10:28 PM] <gdaniels> Srinath: I'm saying we don't actually need SOAP messages to test OM
right away, just XML. But whatever, sure, Alek has a fair point.
[12/1/2004 10:28 PM] <gdaniels> alek: I was agreeing with you - "anything more than that needed?" "no
not really"
[12/1/2004 10:28 PM] <Srinath> geln:but now OM seem to pretty SOAP specific
[12/1/2004 10:28 PM] <alek_s> I am really curious what we would be msising from full SOAP stack ...
[12/1/2004 10:28 PM] <gdaniels> Srinath: No no no
[12/1/2004 10:29 PM] <alek_s> (besides WSDL 1.1 or 2 of course)
[12/1/2004 10:29 PM] <Srinath> but now it is .. it create a SOAPMessage
[12/1/2004 10:29 PM] <gdaniels> Do we have SOAP processing in OM?
[12/1/2004 10:29 PM] <Ajith> ooops Glen : OM is SOAP specific
[12/1/2004 10:29 PM] <gdaniels> oh
[12/1/2004 10:29 PM] <gdaniels> my bad then :)
[12/1/2004 10:29 PM] <Ajith> so wont work for anything other than SOAP messages
[12/1/2004 10:29 PM] <alek_s> lot of people talk about XML messaging over REST so OM could be used in
this context ...
[12/1/2004 10:30 PM] <alek_s> if it could be used as generic XML tool ...
[12/1/2004 10:30 PM] <Chinthaka> yes, OM is using a reduced xml infoset to support SOAP
[12/1/2004 10:30 PM] <gdaniels> That's what I was thinking we were doing
[12/1/2004 10:30 PM] <alek_s> XML Infoset actually
[12/1/2004 10:30 PM] <gdaniels> hm. Clearly I need to go look at the current APIs. :)
[12/1/2004 10:31 PM] <gdaniels> If that's true (SOAP specific) then the modularity is much less compelling
[12/1/2004 10:31 PM] <Srinath> glen now the SOAP OM and OM is one ..
[12/1/2004 10:31 PM] <Ajith> have a look at the prototype2 It has the refactored API
[12/1/2004 10:31 PM] <gdaniels> since it needs to be tightly bound to a SOAP engine to do anything useful
[12/1/2004 10:31 PM] <Ajith>
https://svn.apache.org/repos/asf/webservices/axis/trunk/java/dev/scratch/prototype2/
[12/1/2004 10:32 PM] <gdaniels> updating my SVN tree now
[12/1/2004 10:32 PM] <gdaniels> So why is making it SOAP specific a good idea?
[12/1/2004 10:33 PM] <gdaniels> Are there really a lot of optimizations happening in the code for SOAP?
[12/1/2004 10:33 PM] <Srinath> glen you have spent too much time with computer .. always give status
message .. updating .. reading .. :)
[12/1/2004 10:33 PM] <Ajith> it is very developer friendly :)
[12/1/2004 10:33 PM] <alek_s> from my experience rarely there are any SOAP specific optimizations needed
[12/1/2004 10:33 PM] <Chinthaka> it will give a OM user much user friendly and light weight model and
an API
[12/1/2004 10:34 PM] <Deepal> if it is not SOAP specific , I think we have to suppot full xml inforset
[12/1/2004 10:34 PM] <gdaniels> hm
[12/1/2004 10:34 PM] <alek_s> hmm
[12/1/2004 10:34 PM] <gdaniels> So wait a minute here. We need to support the full XML infoset anyway,
right/
[12/1/2004 10:34 PM] <gdaniels> ?
[12/1/2004 10:35 PM] <Srinath> another status hmm.. :)
[12/1/2004 10:35 PM] <gdaniels> Srinath :)
[12/1/2004 10:35 PM] <Ajith> do we need to?
[12/1/2004 10:35 PM] <Deepal> glen ?
[12/1/2004 10:35 PM] <dims> hang on...am thinking.
[12/1/2004 10:35 PM] <dims> :)
[12/1/2004 10:35 PM] <Ajith> I mean we know PI's are not there in SOAP
[12/1/2004 10:35 PM] <alek_s> SOAP requires almost full Infoset: obnly missing are PIs and Comments and
something aboutNotation
[12/1/2004 10:35 PM] <gdaniels> What about simply echoing a message with no distortion?
[12/1/2004 10:35 PM] <Ajith> and DTS's
[12/1/2004 10:35 PM] <alek_s> thiose are very simpe to support ...
[12/1/2004 10:35 PM] <gdaniels> DTDs
[12/1/2004 10:36 PM] <alek_s> XML Infoset does not require full DTD supports and if there is no DTD in
XML message ....
[12/1/2004 10:36 PM] <gdaniels> yup
[12/1/2004 10:36 PM] <gdaniels> So here's my question. Why not make base OM just about the "raw" XML,
no PIs, maybe no Comments (but I think we should support comments too)
[12/1/2004 10:36 PM] <alek_s> we should just throw exception if you see <![DOCDECL
[12/1/2004 10:36 PM] -->| EC_ (~EC@220.247.221.124) has joined #apache-axis
[12/1/2004 10:36 PM] <gdaniels> Then the SOAP layer sits right on top of that...?
[12/1/2004 10:36 PM] <Ajith> comments are supported :)
[12/1/2004 10:36 PM] <EC_> sorry we missed some stuff here :(
[12/1/2004 10:36 PM] -->| Jaliya5712 (~Miranda@220.247.221.124) has joined #apache-axis
[12/1/2004 10:37 PM] <gdaniels> So OM provides "pseudo-DOM", and SOAP-OM provides "pseudo-SAAJ"
[12/1/2004 10:37 PM] <dims> Here's one...we should be able to load and save all xml's in the
http://www.w3.org/2001/05/xmlschema-test-collection/
[12/1/2004 10:37 PM] <gdaniels> Then you'd be able to use the OM in non-SOAP contexts
[12/1/2004 10:37 PM] <Ajith> We thought it first but then decided it is unwanted complexity
[12/1/2004 10:37 PM] <gdaniels> If there's a good reason to really restrict it to SOAP, then OK. But
I'm not seeing one yet.
[12/1/2004 10:38 PM] <gdaniels> What about allowing Axis2 to support other protocols?
[12/1/2004 10:38 PM] <gdaniels> XOP is not actually SOAP-specific, I believe.
[12/1/2004 10:38 PM] -->| chathurah (~chathura@220.247.221.124) has joined #apache-axis
[12/1/2004 10:38 PM] <Ajith> >
[12/1/2004 10:38 PM] <Srinath> other protocols?
[12/1/2004 10:38 PM] <Ajith> ?
[12/1/2004 10:38 PM] <dims> glen: Let's do both interop client/server and load/save of xmls in the
http://www.w3.org/2001/05/xmlschema-test-collection/
[12/1/2004 10:38 PM] <Deepal> but if OM SOAP sfecific we can easily acess SOAP Hedaders and SOAP body
etc ..
[12/1/2004 10:38 PM] <chathurah> think the *AXIOM* should support the whtever required for Axis so if
Om is used in SOAP processing think we can customise the OM to SOAP
[12/1/2004 10:39 PM] <alek_s> it is possible without much effort to provide SOAP specific API on top of
XML API
[12/1/2004 10:39 PM] <gdaniels> chathura: Isn't it easy to layer the SOAP stuff on top of a more
generic version without losing much?
[12/1/2004 10:39 PM] <alek_s> and also to make it very seemless
[12/1/2004 10:39 PM] <gdaniels> +1 alek
[12/1/2004 10:39 PM] <dims> glen: Let's do both interop client/server and load/save of xmls in the
http://www.w3.org/2001/05/xmlschema-test-collection/
[12/1/2004 10:40 PM] <gdaniels> dims: +1, but I think the current discussion is a little more abstract
[12/1/2004 10:40 PM] <EC_> glen : do we need to make OM, not SOAP specific ?
[12/1/2004 10:40 PM] <gdaniels> I thought we were making OM non-SOAP specific
[12/1/2004 10:40 PM] <gdaniels> That had always been my impression. Then SOAP API would sit on top
[12/1/2004 10:40 PM] <Srinath> making OM not soap specific has plus point of making it easy to test
[12/1/2004 10:40 PM] <dims> that's what i thought too
[12/1/2004 10:40 PM] <Deepal> as chathura said OM is for axis only i guess
[12/1/2004 10:40 PM] <gdaniels> OM would enable things like databinding/MTOM support for generic
(limited) XML
[12/1/2004 10:41 PM] <gdaniels> And SOAP API sits, as Alek said, seamlessly on top of that.
[12/1/2004 10:41 PM] <alek_s> i can provide examples how this can be done
[12/1/2004 10:41 PM] <EC_> I thought OM is to handle SOAP in Axis2 :(
[12/1/2004 10:41 PM] <Srinath> glen do we need to bring the data binding in to OM
[12/1/2004 10:41 PM] <gdaniels> Srinath: I think we do, yes.
[12/1/2004 10:42 PM] <gdaniels> And it's NOT the "data binding" itself. It's *hooks*.
[12/1/2004 10:42 PM] <Srinath> I would like to see OM clean of everything expect XML .. no databinding
..SOAP
[12/1/2004 10:42 PM] <gdaniels> That's all.
[12/1/2004 10:42 PM] <alek_s> i was doing it in farily static way with adapter pattern (easy to debug)
but it is also possible to do it more cutting edge bytecode rewriting
[12/1/2004 10:42 PM] <alek_s> it is possible to "mesh in" SOAP aspect into XML api completely seemless
[12/1/2004 10:42 PM] <gdaniels> bytecode rewriting is super cool but oh-so-hard to debug. :)
[12/1/2004 10:42 PM] <dims> +1 to adapter pattern
[12/1/2004 10:42 PM] <Srinath> yes glen, but It might be better we make pull and oush interafces=hooks
[12/1/2004 10:43 PM] <Srinath> oush=push
[12/1/2004 10:43 PM] <Srinath> alek? cutting edge byt code rewrting
[12/1/2004 10:44 PM] <gdaniels> I gave an example in last week's chat of the kind of API I want to use
as a developer for some databinding stuff.
[12/1/2004 10:44 PM] <gdaniels> Could you do the same thing with how you'd like to do it, Srinath?
[12/1/2004 10:44 PM] <gdaniels> (email is fine)
[12/1/2004 10:44 PM] <EC_> glen : for data binding
[12/1/2004 10:44 PM] <gdaniels> i.e. I'd like to see an example of how you see databinding working with
no hooks for it in OM
[12/1/2004 10:44 PM] <EC_> can we do like this
[12/1/2004 10:45 PM] <EC_> OM exposes pull and push interfaces
[12/1/2004 10:45 PM] <EC_> OM can be build from SAX events
[12/1/2004 10:45 PM] <alek_s> for byte coede meshing use one of free libs (and i think it became even
soemwhat possible in JDK5)
[12/1/2004 10:46 PM] <alek_s> i do not have any favorite one currently and i find them too hard to
debug (at least for now)
[12/1/2004 10:46 PM] |<-- dims has left irc.freenode.net (Remote closed the connection)
[12/1/2004 10:47 PM] <--| chathurah has left #apache-axis
[12/1/2004 10:47 PM] -->| chathurah (~chathura@220.247.221.124) has joined #apache-axis
[12/1/2004 10:47 PM] <EC_> hello
[12/1/2004 10:47 PM] <gdaniels> Hi
[12/1/2004 10:48 PM] <chathurah> :)
[12/1/2004 10:48 PM] <EC_> ahh, something wrong in internet connection here :(
[12/1/2004 10:48 PM] <Deepal> :(
[12/1/2004 10:48 PM] <gdaniels> Did we lose Srinath?
[12/1/2004 10:48 PM] <Srinath> no no:)
[12/1/2004 10:48 PM] -->| dims (~dims@h00045ad8e984.ne.client2.attbi.com) has joined #apache-axis
[12/1/2004 10:48 PM] <Ajith> seems the usual problem when it is the most important hour of the week :)
[12/1/2004 10:48 PM] <gdaniels> Oh there you are :)
[12/1/2004 10:48 PM] <alek_s> hello
[12/1/2004 10:48 PM] <gdaniels> wb dims
[12/1/2004 10:48 PM] <Srinath> procsesing OM ....
[12/1/2004 10:49 PM] <dims> network problems...
[12/1/2004 10:49 PM] <dims> sorry
[12/1/2004 10:49 PM] <gdaniels> So Srinath, what say, could you post a databinding sketch to the list?
[12/1/2004 10:49 PM] <Ajith> for OM?
[12/1/2004 10:49 PM] <Srinath> you mean how to do it with pull and push?glen;
[12/1/2004 10:50 PM] <alek_s> one thing that needs to be clarified is how much OM is between XML stream
and actual XML-Java databinding
[12/1/2004 10:50 PM] <gdaniels> I mean how you envision it looking to the programmer
[12/1/2004 10:50 PM] <gdaniels> I've already said I want stuff like element.addChild(qname, objectValue).
[12/1/2004 10:51 PM] <gdaniels> Obviously that's super-easy for the developer. I want to know how much
work I need to do to make that happen another way.
[12/1/2004 10:51 PM] <gdaniels> OMElement el = new OMElement(qname, objectValue)
[12/1/2004 10:51 PM] <gdaniels> So instead, I'm assuming you'd do something like
[12/1/2004 10:52 PM] =-= EC_ is now known as EChinthaka
[12/1/2004 10:52 PM] <Srinath> glen: I just saying we need to do that thing in a databinding layer on
top of OM
[12/1/2004 10:52 PM] <Ajith> Glen : This may need some work
[12/1/2004 10:52 PM] <gdaniels> OMElement el = databindingFramework.serializeToOM(objectValue)?
[12/1/2004 10:52 PM] <Ajith> Since OM does not have support for that sort of thing
[12/1/2004 10:52 PM] <Ajith> Glen : yeah that could be done
[12/1/2004 10:53 PM] <gdaniels> Ajith: I know it doesn't have support for it now. It doesn't do MTOM
yet either. We could add both.
[12/1/2004 10:53 PM] <EChinthaka> glen : I will propose using an email to list, a databinding framework
on top of OM
[12/1/2004 10:53 PM] <Ajith> I guess what we should do is the second case
[12/1/2004 10:53 PM] <dims> Hang on guys...
[12/1/2004 10:53 PM] <gdaniels> I think we need some use cases
[12/1/2004 10:54 PM] <gdaniels> a) extract java.awt.Image from base64 encoded XML (no MTOM)
[12/1/2004 10:54 PM] <gdaniels> b) extract java.awt.Image from MTOM encoded XML
[12/1/2004 10:54 PM] <gdaniels> etc...
[12/1/2004 10:54 PM] |<-- chathura has left irc.freenode.net (No route to host)
[12/1/2004 10:54 PM] <dims> Let's take an example, xmlbeans, it can consume SAX events or it can pull
STAX events to build the java object. can it work with OM?
[12/1/2004 10:54 PM] |<-- Jaliya has left irc.freenode.net (No route to host)
[12/1/2004 10:54 PM] |<-- Chinthaka has left irc.freenode.net (No route to host)
[12/1/2004 10:54 PM] <EChinthaka> dims : yes
[12/1/2004 10:54 PM] <gdaniels> dims: Yes, because OM can present either
[12/1/2004 10:55 PM] <Srinath> yes Dims.. since OM present pull interce
[12/1/2004 10:55 PM] <alek_s> Glen: as you know i really like to talk about test/use cases :)
[12/1/2004 10:55 PM] <gdaniels> it can push SAX events for a given element, or it can be asked for StaX
events
[12/1/2004 10:55 PM] <dims> So if people can use Castor or XMLBeans or JAXB....you see where i am going?
[12/1/2004 10:55 PM] <Srinath> I thought as we have stax on om we do not ned anything else
[12/1/2004 10:55 PM] <dims> for O1/M1
[12/1/2004 10:56 PM] <alek_s> Dims: if people directly access XML event streams (push or pull) they do
not need OM
[12/1/2004 10:56 PM] <EChinthaka> alek : in that case OM can expose the raw pull stream
[12/1/2004 10:56 PM] <Srinath> alek they do .. if they want to add handlers
[12/1/2004 10:56 PM] <alek_s> Dims: but they will also not get SoapEnvelope and AXIS2
header/chans/MTOM/other goodies
[12/1/2004 10:56 PM] <gdaniels> alek : -1. They still need OM to some extent because Axis will be
doing the work on the SOAP header and providing access through the body
[12/1/2004 10:57 PM] <EChinthaka> but thru OM
[12/1/2004 10:57 PM] <alek_s> that is where i was going :)
[12/1/2004 10:57 PM] <alek_s> OM is to bind them all :)
[12/1/2004 10:57 PM] <gdaniels> dims: Where were you going?
[12/1/2004 10:57 PM] <dims> alek: idea is axis will use OM better than these others
[12/1/2004 10:57 PM] <alek_s> LOTR anynone ...
[12/1/2004 10:57 PM] <gdaniels> oh you mean let the databinding work on the XML events...yeah yeah
[12/1/2004 10:57 PM] <gdaniels> and what of MTOM?
[12/1/2004 10:58 PM] <Srinath> +1 for databinding work on XML events
[12/1/2004 10:58 PM] <dims> that makes 3 items on my list...interop client/server, Load/Save XML in the
Schema url above, data-binding samples
[12/1/2004 10:58 PM] <dims> for O1/M1
[12/1/2004 10:58 PM] <dims> for XMLBeans/Castor/JAXB
[12/1/2004 10:59 PM] <gdaniels> I'm not sure we need databinding samples for O1
[12/1/2004 10:59 PM] <gdaniels> just enough between us that we all agree it can be done efficiently and
easily
[12/1/2004 10:59 PM] <Srinath> glen, dims are we agreeing on something :)
[12/1/2004 11:00 PM] <dims> samples never hurt :)
[12/1/2004 11:00 PM] <Srinath> yap
[12/1/2004 11:00 PM] <gdaniels> sure
[12/1/2004 11:00 PM] <dims> proof that we can use existing databinding stuff even with this early prototype
[12/1/2004 11:00 PM] <dims> just a few :)
[12/1/2004 11:00 PM] <gdaniels> hokay
[12/1/2004 11:00 PM] <dims> thanks.
[12/1/2004 11:01 PM] <alek_s> +1 to samples
[12/1/2004 11:01 PM] <dims> +1 to test cases :)
[12/1/2004 11:01 PM] <gdaniels> I want Axis2 to be at least as easy to use for a Java programmer doing
basic data binding as Axis1. That's my goal.
[12/1/2004 11:01 PM] <gdaniels> As long as we get there I'm happy.
[12/1/2004 11:01 PM] <dims> amen
[12/1/2004 11:01 PM] <Srinath> +1 .. shall we restate what we agreed .. o make thing clear
[12/1/2004 11:01 PM] <alek_s> :)
[12/1/2004 11:01 PM] <gdaniels> If I'm forced to send XML pull streams to a databinding framework, I
won't be happy....
[12/1/2004 11:02 PM] <gdaniels> (long term - fine for O1)
[12/1/2004 11:02 PM] <Srinath> glen we can provide utilities to make the developers life easy
[12/1/2004 11:03 PM] <Srinath> then all are happy
[12/1/2004 11:03 PM] <gdaniels> Srinath: +1, I just want to see what it ends up looking like
[12/1/2004 11:03 PM] <dims> let's see how O1 shapes up
[12/1/2004 11:03 PM] <Srinath> +1 :):)
[12/1/2004 11:03 PM] <gdaniels> OK, so let's take this to the mailing list / wiki
[12/1/2004 11:03 PM] <gdaniels> We need to decide exactly what goes in O1
[12/1/2004 11:03 PM] <farhaan> If we are agreeing to something I would suggest that for everybody's
benefit we update it on the OM Requirements/use cases page. This removes ambiguity and can be referred back
[12/1/2004 11:03 PM] <alek_s> utilities of that kind should be part of API as they are used in majority
of cases
[12/1/2004 11:04 PM] <alek_s> +1 Farhaan
[12/1/2004 11:04 PM] <gdaniels> alek: I agree with you. The rest of these guys will take some
convincing. :)
[12/1/2004 11:04 PM] <dims> Please document a Release plan for O1 in wiki
[12/1/2004 11:04 PM] <gdaniels> Who wants to take an AI to get an outline for that going?
[12/1/2004 11:05 PM] <farhaan> AI?
[12/1/2004 11:05 PM] <gdaniels> Action Item
[12/1/2004 11:05 PM] <farhaan> I will do that
[12/1/2004 11:05 PM] <Srinath> or Artifical inteligance ;)
[12/1/2004 11:05 PM] <gdaniels> a trackable "todo"
[12/1/2004 11:06 PM] <gdaniels> :) Srinath
[12/1/2004 11:06 PM] <farhaan> http://wiki.apache.org/ws/FrontPage/Axis2/ProjectPlan
[12/1/2004 11:07 PM] <farhaan> definitely we need to discuss each item that was put down on Sanjiva's mail
[12/1/2004 11:07 PM] <gdaniels> No offense, guys, but you also need to work on your spelling a little.
Misspelled class names are not acceptable, IMO....
[12/1/2004 11:08 PM] <dims> +10000
[12/1/2004 11:08 PM] <gdaniels> s/Manger/Manager/, etc... :)
[12/1/2004 11:08 PM] <Ajith> Sure we will correct whatever the mistakes soon
[12/1/2004 11:08 PM] <Srinath> will lookin to that
[12/1/2004 11:08 PM] <gdaniels> thx!
[12/1/2004 11:08 PM] <farhaan> definitely
[12/1/2004 11:08 PM] <Jaliya5712> We will take necessary actions to avoid that
[12/1/2004 11:08 PM] <gdaniels> I know I'd do worse if I were coding in Sinhalese... :)
[12/1/2004 11:08 PM] <Jaliya5712> I mean check all the classes :)
[12/1/2004 11:09 PM] <chathurah> +1
[12/1/2004 11:09 PM] <Srinath> I am looking for a IDE with spell checker :D
[12/1/2004 11:09 PM] <EChinthaka> :D
[12/1/2004 11:09 PM] <Jaliya5712> eclipse has plugin
[12/1/2004 11:09 PM] <chathurah> Jaliya found a one:d
[12/1/2004 11:09 PM] <alek_s> (side thought: how many AIs does it take to make AXIS2?)
[12/1/2004 11:09 PM] <dims> :)
[12/1/2004 11:09 PM] <gdaniels> Oh, we still have an open issue about generic XML vs. SOAP specific OM
[12/1/2004 11:10 PM] <Srinath> jaliya are you serious?
[12/1/2004 11:10 PM] <gdaniels> That needs to get decided
[12/1/2004 11:10 PM] <EChinthaka> are we thinking of using OM for non-SOAP xmls ?
[12/1/2004 11:10 PM] <gdaniels> alek: "IRC - putting the 'AI' in 'AXIS2'..."
[12/1/2004 11:10 PM] <Jaliya5712> Srinath:Yes, I will send them
[12/1/2004 11:11 PM] <farhaan> yes Glen, but do we have to make it XML specific since we are developing
a SOAP engine
[12/1/2004 11:11 PM] <alek_s> AXIS will rule: it has more than one AI inside ...
[12/1/2004 11:11 PM] <chathurah> btw i wanted to discuss the Service Desc functionality of the WSDL
Object Model
[12/1/2004 11:11 PM] <Ajith> Glen : have allok at the interface and let us know the things that you
need improvements on
[12/1/2004 11:11 PM] <gdaniels> EChinthaka: It was my impression (and apparently dims' too, and perhaps
Alek's too) that we were making OM generic with SOAP on top of that
[12/1/2004 11:11 PM] <EChinthaka> glen : ok
[12/1/2004 11:11 PM] <alek_s> that is what i always thought - generic and adaptable to put SAAJ, DOM,
etc on top of it ...
[12/1/2004 11:11 PM] <gdaniels> If that's not the case, and we're going to shut out non-SOAP uses of
OM, I want to make sure there is a really good reason for it.
[12/1/2004 11:12 PM] <gdaniels> Me too.
[12/1/2004 11:13 PM] <gdaniels> I'll take a look at the current API in proto2 and send a proposal, OK?
[12/1/2004 11:13 PM] <dims> +1
[12/1/2004 11:13 PM] <Ajith> thats better
[12/1/2004 11:13 PM] <Deepal> Glen : Thats better
[12/1/2004 11:13 PM] <gdaniels> Alek, I'll probably ping you to help out
[12/1/2004 11:14 PM] <Srinath> +1 glen
[12/1/2004 11:15 PM] <gdaniels> ok, I need to get to bed folks - early meeting tomorrow.
[12/1/2004 11:15 PM] <alek_s> ok glen
[12/1/2004 11:15 PM] <dims> bye all
[12/1/2004 11:15 PM] <Deepal> bye dims
[12/1/2004 11:15 PM] <farhaan> bye
[12/1/2004 11:15 PM] <alek_s> i will post itc log
[12/1/2004 11:15 PM] <gdaniels> good night, everyone!
[12/1/2004 11:15 PM] <EChinthaka> bye alll
[12/1/2004 11:15 PM] <gdaniels> send more email
[12/1/2004 11:15 PM] <alek_s> bye everybody
[12/1/2004 11:15 PM] <gdaniels> thanks alek
[12/1/2004 11:15 PM] <Ajith> byee
[12/1/2004 11:15 PM] <chathurah> bye