synapse-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Paul Fremantle <>
Subject chat log
Date Thu, 22 Dec 2005 14:05:49 GMT
[13:00] saminda: Hi folks
[13:00] paulfremantle: hi!
[13:01] saminda: Paul i have replied to your mail, what would you think
[13:02] paulfremantle: let me take a look
[13:02] saminda: once other guys are connected, pls let us know the current
state of Synapse
[13:02] paulfremantle: saminda
[13:02] paulfremantle: im not sure i agree that you should get an internal
server error on a fault
[13:03] saminda: for ex: release plan and how we should have a good test
[13:03] paulfremantle: ok
[13:04] saminda: Paul, what should initiator expect on <fault/>
[13:04] saminda: I've done major changes to the current code, pls be kind
enough to look at them too
[13:05] saminda: lot of hacks are there since axis2 folks fix sum bugs :)
[13:06] *** pvikas has joined #apache-synapse.
[13:07] paulfremantle: hi vikas
[13:07] saminda: Hi Viks
[13:07] pvikas: hi
[13:08] pvikas: so we wait for ppl ...
[13:08] paulfremantle: i know ant is trying to get on
[13:08] saminda: Vikas you promised a Load balancing sample
[13:09] paulfremantle: that would be COOL
[13:10] pvikas: am working. but i need a sendMediator and also the thing
thats on the mailing list
[13:10] pvikas: return value and the to combo
[13:10] paulfremantle: by the way saminda... i'm looking at the SOAP Fault
return codes and i think 500 is a good default value
[13:11] saminda: cant you do it with <servicemediator/> Vikas
[13:11] paulfremantle: i think we ought to be able to set it tho using
parameters in the <fault> tag
[13:11] paulfremantle: e.g. <fault rc="500">
[13:11] pvikas: failover would need the actual status as returned by the
[13:11] pvikas: no defaulting or civerups by synapse
[13:11] paulfremantle:  <faultcode>blah</faultcode>
[13:11] paulfremantle: etc
[13:11] pvikas: oops.. coverups
[13:11] saminda: failour need 4xx/5xx
[13:12] saminda: to do the job
[13:12] *** ant has joined #apache-synapse.
[13:12] paulfremantle: i see what you mean vikas
[13:12] paulfremantle: hi ant
[13:12] pvikas: [saminda] the sendMediator would replicate send Processor?
[13:12] ant: hi, sorry, bit late
[13:12] pvikas: hi ant
[13:12] pvikas: it would depend on sendBack/send On
[13:14] saminda: this what exactly Pauls new mail says
[13:14] pvikas: i tried putting a JIRA for am going through that
[13:15] saminda: SendOnProcessorWithRuleTest check it out
[13:15] pvikas: oops, tried the JIRA for sendOn, am going through paul's
[13:15] saminda: it's based on your scenario
[13:16] saminda: paul, so <falut/> will send a SOAP fault right ?
[13:16] paulfremantle: yes
[13:16] paulfremantle: well <fault/> will
[13:16] pvikas: the current thing sends an exception in case the mediator
returns false...
[13:17] paulfremantle: and <fault
[13:17] saminda: vikas, the SendOnProcessorWithRuleTest will use the
sampleRouting mediator
[13:17] paulfremantle: does it?
[13:17] saminda: it is the switch between sendOn and sendBack
[13:18] saminda: paul, +1
[13:19] saminda: vikas. we haven't put a test case to test mediator return
[13:19] saminda: yet
[13:19] pvikas: ok
[13:19] saminda: that's why we need comprehensive test sutie to test all the
possible scenarios
[13:19] saminda: will put a test case asap
[13:20] saminda: as you mention it synpase throws an exception when mediator
retruns false
[13:20] saminda: we need to fix it
[13:21] paulfremantle: that is definitely a bug
[13:21] pvikas: Synapse should always return a reply, maybe with a soap
fault / custom message
[13:21] paulfremantle: no i dont agree vikas
[13:21] paulfremantle: what if the message is a Denial of Service attack
[13:21] paulfremantle: then maybe i just want to drop the connection
[13:22] saminda: Synapse doesn;t have to return a message, but it should
return some
[13:22] pvikas: and how would you diff between original  demand for a
popular service and a DoS attack?
[13:22] saminda: http status
[13:22] saminda: like 202 ok
[13:23] paulfremantle: its up to the mediators to decide whether to drop,
send a fault or send a response
[13:23] saminda: paul what will be the work around for DOS
[13:23] saminda: sending a fault for DOS is more appropriate, what would you
[13:23] paulfremantle: but if a mediator says drop then there should be just
a 200OK or a 202OK or a 500 HTTP close
[13:23] saminda: think
[13:23] paulfremantle: i think its up to whoever configures synapse
[13:24] paulfremantle: but i dont think the engine should assume there is
always a soap response
[13:24] saminda: yes, i agree, more flexibility
[13:24] pvikas: and how will you diff detween a DoS and a stress/ perf
test.. the mediator would drop the connection, bad results..
[13:24] saminda: yes REST is always possible
[13:25] paulfremantle: vikas.... its not up to me to decide
[13:25] paulfremantle: its up to whoever writes the mediator that looks for
[13:25] paulfremantle: but if we always return a result
[13:25] paulfremantle: then no-one can write a mediator that doesnt return a
[13:26] saminda: +1 paul
[13:27] saminda: there is http status, "bad request"
[13:27] pvikas: [confused].. A security mediator says you can not go ahead,
returns false ... what happens?
[13:27] saminda: 403 may be, can be used in DOS
[13:27] paulfremantle: vikas
[13:27] saminda: stop processing
[13:27] pvikas: yup
[13:27] paulfremantle: who writes the mediator decides
[13:27] paulfremantle: they can either return a fault
[13:27] paulfremantle: or simply an HTTP code
[13:28] saminda: send 5XX
[13:28] paulfremantle: the framework doesn't force an answer
[13:28] pvikas: well, the mediator actually decides the response to be sent
[13:29] saminda: {Vikas do you have a Jira account }
[13:29] pvikas: ya, i do..
[13:29] paulfremantle: now if we ship a security mediator, I suggest we send
back 401 Unauthorized + a soap fault
[13:29] saminda: {when you log in can you see the create issue}
[13:30] pvikas: It shows this when i try to file up an issue
[13:30] pvikas: Errors
[13:30] pvikas: Assign To: The default assignees do NOT have ASSIGNABLE
permission OR Unassigned issues are turned off.
[13:30] pvikas: ok, paul
[13:31] saminda: +1 Paul, but Axis2 doesn't have a way to handle 4xx
[13:31] saminda: we use axis2's tranport
[13:31] saminda: are we thinking of our own transport :)
[13:31] paulfremantle: NO
[13:31] pvikas: no
[13:31] paulfremantle: i suggest we go with what axis2 supports
[13:32] saminda: they pls be satisfied with 500 :)
[13:32] paulfremantle: if we need other HTTP rcs then we can add that to
[13:32] saminda: oh...
[13:32] paulfremantle: ok i will be happy with 500 for now
[13:32] pvikas: i will live too
[13:33] paulfremantle: right
[13:33] paulfremantle: Milestone1 release
[13:33] paulfremantle: I have asked the WS-PMC to approve it
[13:33] paulfremantle: then I will ask the Incubator PMC
[13:33] saminda: is it going to be Synpse-M1-RC1 :)
[13:33] saminda: has it
[13:34] paulfremantle: it has to be
[13:35] paulfremantle: synapse-incubating-m1-rc1
[13:35] paulfremantle: we must include the term incubating in the filename
[13:35] paulfremantle: saminda can you please update maven to make that
[13:36] pvikas: [doubt] m= milestone, rc??
[13:36] paulfremantle: m = milestone, rc = release candidate
[13:36] pvikas: thanks
[13:37] pvikas: the build is breaking,
[13:37] saminda: sure paul, it's just one line of update :)
[13:37] paulfremantle: ah well we need to fix that
[13:37] paulfremantle: :)
[13:38] pvikas:     [javac]
[13:38] pvikas: 114: getServiceContext(
org.apache.axis2.description.AxisService) in
[13:38] saminda: Build is not breaking
[13:38] pvikas: GroupContext cannot be applied to (java.lang.String)
[13:38] pvikas:     [javac]                     ServiceContext sc =
[13:38] *** Sathish has signed off IRC ("ChatZilla 0.9.61 [Mozilla rv:1.7.2
[13:39] saminda: you should have the latest axis2-0.94-SNAPSHOT in you IDE
[13:39] saminda: classpath
[13:39] pvikas: i just got the synapse source code, tried maven dist-bin
[13:39] paulfremantle: doesn't maven do that for you saminda?
[13:39] pvikas: it does
[13:40] pvikas: and samina please try builds without the IDE [ u do remember
the 'classes' issue ?? ]
[13:40] saminda: ProxyStockQuoteClient doen't have to do anything on build
[13:40] saminda: samples are passing
[13:41] paulfremantle: im just building without my ide to see
[13:41] saminda: see these days we have done so many changes to axis2
[13:41] saminda: it should compile, since this is runtime we need to fix the
code in the clinet
[13:42] pvikas: ah.. i am doing a build, not a recompile of the client
[13:42] saminda: Vikas i have a seperate copy to see the build :)
[13:42] pvikas: cool
[13:43] saminda: buid fail means one of you test case is failing or a
comilation or no resources to copy
[13:43] pvikas: compilation is what i had pasted
[13:43] saminda: wait
[13:44] *** ant has signed off IRC (Read error: 110 (Connection timed out)).
[13:46] paulfremantle: im getting same build failure saminda
[13:46] pvikas: +1
[13:46] saminda: paul are you trying the build with test skip
[13:48] saminda: Pual i have a seperate copy in my system to see if i'm
mistakenly forgotten
[13:48] saminda: somthing to commit
[13:48] saminda: i tried that one now, with the latest update, i'm not seen
a build fail
[13:49] paulfremantle: im just trying to fix it
[13:49] paulfremantle: its another change in axis2
[13:49] saminda: axis2 changing every min
[13:50] saminda: it's very hard to keep on it
[13:50] saminda: but i'm doing my best
[13:50] paulfremantle: im just checking in the fix
[13:51] saminda: thanks paul
[13:52] saminda: what is the status of test suite
[13:52] saminda: we need all the help we can get
[13:52] paulfremantle: yes we need more tests
[13:52] paulfremantle: we have very patchy coverage
[13:52] saminda: Vikas found a problme if mediator return false it sends
exception that should be fixed
[13:53] pvikas: also me and saminda are finding things in axis2 too
[13:54] saminda: todo list paul
[13:54] paulfremantle: can we try to each add 3 tests
[13:54] paulfremantle: and also each do some unit test
[13:54] saminda: :) i've added a lot
[13:54] paulfremantle: before tomorrwo
[13:54] paulfremantle: i mean system test
[13:55] paulfremantle: and then we will make it the release candidate
[13:55] paulfremantle: and then we can have a relaxed christmas
[13:55] saminda: ho ho ho....
[13:55] paulfremantle: i think apart from tests we have almost everything we
need for m1
[13:56] paulfremantle: i will add to the post m1 todo list
[13:56] pvikas: yup we do..
[13:56] saminda: ok will do
[13:56] paulfremantle: saminda the security test is breaking the build
[13:56] saminda: when you test skip right
[13:56] paulfremantle: can we please take security out of M1
[13:56] saminda: i will fix it
[13:57] paulfremantle: since it wasnt in the plan and move it to M2
[13:57] saminda: sure
[13:57] paulfremantle: can you make it a new directory and use the extension
[13:57] paulfremantle: so then we can leave it in the head but move it out
of maven
[13:57] paulfremantle: in future
[13:57] saminda: will do it, it's not completed yet
[13:57] paulfremantle: can we agree a model
[13:57] paulfremantle: where we put new stuff
[13:57] paulfremantle: into scratch
[13:58] saminda: ok
[13:58] pvikas: better done in the mailing list, only the 3 of us are here..
[13:58] paulfremantle: and get it really good... ask people to review it...
and then check into HEAD
[13:58] paulfremantle: yes vikas
[13:58] paulfremantle: do you think thats a good model?
[13:58] saminda: will put to scratch and clean it up
[13:58] pvikas: i dont mind the location, things move into main-trunk only
if ppl agree
[13:58] saminda: i have put hacks till some bugs fix in axis2, is that
[13:58] paulfremantle: ok cool
[13:58] paulfremantle: yes
[13:59] paulfremantle: but we should wait until there is a stable .94 before
we release
[13:59] saminda: it may be .9652
[13:59] pvikas: +1 paul, i think we should try and synchronize with axis2 at
intervals and not always
[14:00] paulfremantle: cool +1 to .9652
[14:00] paulfremantle: vikas +1
[14:01] paulfremantle: we had agreed to stick to .93 but then at the
hackathon we came across some "showstoppers"
[14:01] saminda: yep, proxies :)
[14:01] pvikas: what about the synapse hackathon, nothing on the list?
[14:01] paulfremantle: in general we should try to stick to the latest
stable release (or previous) and only move when there is a new stable
[14:01] paulfremantle: well we mainly just did minor stuff
[14:02] paulfremantle: and discussed some of the things i have posted on
[14:02] pvikas: ur presentation was cool and covered good stuff
[14:03] paulfremantle: thanks
[14:03] paulfremantle: it went well
[14:03] paulfremantle: and it was fun
[14:03] paulfremantle: maybe next year you will make it over vikas?
[14:04] paulfremantle: right i have to go
[14:04] pvikas: will try to, too many Visa restrictions here
[14:04] paulfremantle: yes i know its hard
[14:04] pvikas: happy christmas
[14:04] paulfremantle: maybe apachecon europe will be easier?
[14:04] pvikas: and a great weekend
[14:04] pvikas: who is doing the log?
[14:04] paulfremantle: yes have a great christmas
[14:04] saminda: I have to go too, Happy Christmas
[14:04] pvikas: paul?
[14:04] paulfremantle: i probably wont be here next week
Paul Fremantle
VP/Technology, WSO2 and OASIS WS-RX TC Co-chair

"Oxygenating the Web Service Platform",

View raw message