Interoperability Minutes 2009-04-20

From MemberWiki

Jump to: navigation, search

URL: http://www.openajax.org/member/wiki/Interoperability_Minutes_2009-04-20

Contents

Attendees

  • Jon Ferraiolo, IBM
  • Matthias Hertel
  • Javier Pedemonte, IBM
  • Howard Weingram, TIBCO

Original Agenda

Minutes

Issue 1: version string in iframe protocol

Jon: Is this a spec or an implementation issue?

Howard: Implementation. Not a high priority for any of us. I have other things to say on this issue but haven't sent mail.

Javier: I've received feedback from people at IBM. They are worried about future updates and hosts and clients from different software providers

Howard: I suggested protocol id includes major version number. Minor version number is separate. And say that we don't expect compatibility code for major version differences. And for minor should attempt to support other versions. But trickiness is which side decides. Container starts the conversation. HubClient could then talk through tunnel to ask about older versions. But lots of tricky bits, such as redirect issue. IMO, the reference implementation doesn't have to do anything about this within this version.

Jon: There is a possibility in the future of adding a 3rd parameter, params, to the connect() API.

Howard: Another option would be to add more information to getPartnerInfo.

Matthias: Protocol defines how data is exchanged. Other information is about the implementation of the Hub.

Howard: Yes. I want to make sure we are putting version numbers on the interface.

Jon: Just do declarative for now via URL param.

Howard: Yes, only say refimpl supports on protocol now.

Matthias: Yes, good idea.

Issue 2: Subscriber invocation order: Hub2.0 versus Hub1.0, Issue 3: Hub 1.0 test suite issues

(resolved in email, reconfirmed during phone call)

RESOLUTION: Reconfirmed that filterFunc and reinit() are deprecated. Need to note that in the spec. Can remove filter tests from the test suite.

Issue 4: Hub 2.0 charset issues with iframes

Jon: Per Howard's email, we use JSON for transmitting payloads, which requires Unicode and "defaults" to UTF-8. We should say that people SHOULD use UTF-8 for data that will end up in a message

Matthias: Agree

Jon: But what exactly goes in the spec, and where to put it? Based on emails, you can't just say you must use UTF-8.

Howard: Inline cases don't require UTF-8. It is much safer for implementations to use UTF-8. Allows mashups to work across browsers and in combination with things from other parts of the world

Jon: If not UTF-8, should put content in an iframe and convert payloads to UTF-8

Howard: Yes, but challenging to do the conversion

Jon/Howard: SHOULD use UTF-8 so that compatible with content from around the world

Matthias/Howard: encodeURL on ECMA says UTF-8 encoding

Matthias: We should specify UTF-8 but restrict to ASCII 127 to give change to work with other code pages

Jon/Howard: Yes

Jon: Haven't done anything out of range yet

RESOLUTION: SHOULD use UTF-8 so that compatible with content from around the world

Issue 5: iframe clients errors due to browser caching

Howard: I need to look at this still

Javier: I see it only on Firefox, but have workaround code for now, which is to set 'src' after putting the IFRAME into the DOM. But IE6 gives error on HTTPS if no 'src', so put a dummy 'src' on first. I included a long comment in the code.

Issue 6: Sequencing, FIM and the back button

Javier: window.location.replace has no history impact (versus assignment). And as a safety factor, FIM has a sequence counter.

Howard: I need to run my tests again. Some tests were back-button sensitive

getParameters

Howard: I requested this

Javier: No problem

Howard: I added that to the page

Javier: I'll fix the code

Test suite

Jon: Very far along now. Javier and I have added lots of test cases covering nearly all of the APIs. Nearly done. There are about 10 different test case files each with many tests inside.

Howard: Have you covered all scenarios from my wiki page?

Javier: Most of them. No integration tests yet. Anything we need to cover there?

Howard: Different types of integrations. Managed Hub nesting.

Javier: Already have nesting.

Jon/Howard: Add some dynamic tests where calls subscribe and unsubscribe.

(discussion about looping)

Howard: I had a check in my code to look at nesting level. 200 callbacks deep max. I don't remember if I submitted it.

Build scripts

Jon: We have build scripts. Invoke shrinksafe on the JS.

Howard: Nice

Closing out Hub 2.0

Javier: Left to do: getParameters, protocol, and test suite. Can we close down this week?

Jon: Yes on APIs, code and test suite. But voting takes a number of weeks.

Howard: We still need to clean up the spec. Still some red-colored notes.

Jon: We can take a snapshot and call it build 288 or whatever

Howard/Javier: Call it our release candidate

Retrieved from "http://www.openajax.org/member/wiki/Interoperability_Minutes_2009-04-20"
Personal tools