[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: ICE - Messaging, Retries, Sotherby's Auctions, eBAY and Reuters News
OK, I just read these interconnected threads. I'm with Marty on this one! Actually I raised this with Monica and Dale on Thursday too on our call together. I'm also with JJ - I'd like to get rid of this flag - its very confusing as Steve has already shown. I think if people want to do this kinda stuff - they should be using status tracking - eg aka BCM Appendix B- Linking and Switching spec' and we have tabled this feature set for V3. If we keep the flag - then we need to rename it at the BTA level - to avoid all confusion with transport retries. And then - I'm convinced endsWhen / beginsWhen and variables do a much better job anyway - so I'd vote for removing this flag still! Here's my perspective. We started with EDI - and there it is clear. Any kinda payload hanky-panky is called 'certification testing'. You do not switch to production until payloads pass your business rules. This is where CAM is a wonderful tool - since it can pre-validate content formats - prior to them being sent 'live'. Now if rules change in the middle of production runs (because you hit some old legacy date you forgot about!) - that is called versioning and trading partner context - again - see CAM - you handle that by rolling new rule files and doing a production update - or if you are Walmart - telling the grotty little partner to fix his data! This is why you send people transaction Reject messages like 997's with specific application error messages in them. Next - the ICE stuff really went into all this RETRY scheduling in great detail - and now they have version 2 out. Basically its Bruce Hunts teams ideas of the syndication model - eg participants (this is multi-party collaboration - with ocassionally available servers) setup windows and sliding time frames when they will do interchanges - but this has now morphed into the ICE V2 model and SOAP. This works great for auction houses like Sotherby's and eBay - where they collect data, and then subscribers come in and get it - this is also the Reuters News model too. see - http://www.w3.org/TR/NOTE-ice for the intractable W3C mumbo-jumbo, and more clearly this- http://www.icestandard.org/resources/Laird_Popkin.pdf and technical details - http://www.icestandard.org/resources/tech_review.pdf Basically WfMC and RosettaNet stole this ICE stuff blind and re-purposed it. So here we are again - full-circle. My take on this is that BPSS V2 is not where we decide how to handle these syndication models and reties!! Let's table this to V3. Since ICE is already available for people wanting to do this - maybe we should be looking at how ICE and BPSS work together....? If we keep this flag (I know how Monica frets this stuff!) - it should definately be renamed IMHO. Eh - note to Steve Capell - Steve - now you know why I waited off replying to your original post, I don't jump into just any old billabong just because its there; there could be a nasty old croc at the bottom, eh?! Cheers, DW p.s. found this by accident searching for ICE stuff: - http://dev.zope.org/Members/jim/JimsWiki/XMLProtocolComparison/view
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]