Greetings ITC Techies, Another issue that became apparent at the face-to-face and today is that we need to develop a consistent terminology for what is a transaction, what is an element, what is a fragment, what is a thing marked by EOM, etc. I would like to propose the following terminology to be used. I'm not claiming this is universally correct terminology. In fact I don't think there is a universally correct definition of a transaction. But, within the realm of the SCE-MI world, let's at least use consistent terminology and something upward compatible with SCE-MI 1. Transaction: I would like to propose that we define transaction generally as "the amount of data" passed in a single DPI function call. This definition has generally served us well since SCE-MI 1, and all the way through plain DPI. Where it becomes fuzzy is when data shaping and VLM become involved. As we know with data shaping, elements of transactions can be passed in single calls at one (narrow) end of a nozzle or funnel. So, with that in mind, let's qualify the definition of transaction above as above as "the data passed in a single call of a DPI function, or in the case of a nozzle or funnel transaction pipe, the data passed at the larger end of that pipe." Element: That segues us nicely into the definition of an element as simply the data passed at the smaller end of a nozzle or funnel. In other words 1 or more elements of a transaction can be passed in single calls to transaction pipes. Frame: Let's refer a sequence of transactions marked by an eom as a "frame". Transmissions of frames require multiple calls that pass a transaction in each call to conventional pipes (both ends same size). Or to the larger end of a data shaping pipes in the sense of a nozzle or funnel. So, with this in mind, the general granularity of data organization in the SCE-MI 2 context from coarsest to finest is: Frames contain one or more Transactions contain one of more Elements And, except for the special case of elements used in data shaping, transactions are defined in what is passed in a single call to a pipe. How does this sound ? I think it scales well to more traditional definitions of transactions in the SCE-MI 1 and plain DPI senses. -- johnS ______________________________/\/ \ \ John Stickley \ \ \ Principal Engineer \ \________________ Mentor Graphics - MED \_ ________________________________________________________________Received on Thu Mar 2 11:49:59 2006
This archive was generated by hypermail 2.1.8 : Thu Mar 02 2006 - 11:51:01 PST