Paul Clarke wrote: > Hi Tim, > Your point on the number of interfaces is quite correct, however I think > the formula should be: > > For n systems, where n> 1, the number of interfaces is : n(n-1)/2 > e.g. 4 systems requires 6 interfaces, 5 systems = 10 interfaces, 6 > systems = 15 interfaces, 10 systems require 45 interfaces .... and then > we move into mind blowing numbers !!
Whether you need to divide by two depends, of course, on how you define an "interface" - but you are right in that bi-directionality is usually assumed. Alas, the transformation of health data is rarely perfectly symmetrical or commutative, so although the effort in writing those interfaces involved is probably less than n(n-1), but is usually more than n(n-1)/2 Tim C > Tim Churches wrote: > >> David Guest wrote: >> >> >>> Webservices does not seem that daunting a technology. My concern is >>> about standardising the minimum datasets and calls that they will >>> provide. Communication between disparate applications is much easier >>> with RPC/SOAP but you still don't want to write (n-1)! interfaces. >>> >> >> It's not as bad as you think, David! To get n information systems which >> each use their own dataset definitions and semantics to talk to one >> another, you only need write n(n-1) sets of interfaces, don't you?. For >> n > 4, n(n-1) < (n-1)! >> >> However, either way you are on a hiding to nothing. Bring on the >> minimum/core dataset definitions (or I should say, let the argy-bargy >> over what needs to be in the minimum/care dataset definitions commence!)! >> >> Tim C >> >> >> > _______________________________________________ Gpcg_talk mailing list [email protected] http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk
