OASIS Web Services Interactive Applications TC

 View Only

RE: [wsrp] [wsia] [wsrp-wsia joint interfaces] Search results pag e,map services, transient entities.

  • 1.  RE: [wsrp] [wsia] [wsrp-wsia joint interfaces] Search results pag e,map services, transient entities.

    Posted 06-13-2002 05:33
    You're on the spot, for the OO "camp" thinking. In the non-OO "camp" way of thinking, the transient entity is not per-user and thus cannot store the (hotel=DownTownHilton...) information, and this must be stored in the session. But the session is shared (per-user) for all those hotel services, and so the information there must be "partitioned" by the transient entity handle, e.g. the session will be "#1(hotel=hilton...),#2(hotel=wyndham...),#3...", where #n is the transient entity handle. Note: 1. Usually, the information itself will not be stored in the session handle, but rather a handle will be stored which points to the information, but it still will be partitioned, e.g "#1(#123), #2(#456),..." 2. It is the producer/entity that must do the partitioning, and given that an entity can never know whether it can be added multiple times to the same page, it must always do so, or never support the ability to add itself multiple times to the same page. This makes it very unnatural for a "web-oriented" developer to develop a service (they _never_ dealt with this concept before WSIA/WSRP service development), and so I question the phrase "web-architecture group" for the non-OO camp. For them, paradoxically, the OO camp concepts would be more natural, as it would enable them to map _their_ concept of session to the WSIA/WSRP concept of transient entity handle. 3. This email is _not_ an endorsement of any "camp". I see the two sides as "simplicity of model/concept" (the OO camp) vs. "scalability" (the non-OO camp, although a previous email of mine questioned that). Gil