[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: Proposal for 'shared' in a TA Set
i.e. <shared> <normativeSource> <comment>comment2</comment> <interpretation>interpretation2</interpretation> <refSourceItem>refSourceItem2</refSourceItem> <refSourceItem>refSourceItem3</refSourceItem> <textSourceItem>textSourceItem2</textSourceItem> <textSourceItem>textSourceItem3</textSourceItem> <derivedSourceItem>derivedSourceItem2</derivedSourceItem> <derivedSourceItem>derivedSourceItem3</derivedSourceItem> </normativeSource> <target>target0</target> <predicate>predicate0</predicate> <prerequisite>prerequisite0</prerequisite> <prescription>prescription0</prescription> <comment>comment1</comment> <interpretation>interpretation1</interpretation> <var>var0</var> <var>var1</var> <tag>tag0</tag> <tag>tag1</tag> </shared> Best regards --- Stephen D Green 2009/10/30 Stephen Green <stephen.green@documentengineeringservices.com>: > I start with a premise that there is a single default intention > regarding concurrency behaviour for each of the TA parts > when it is shared across a TA Set. > > > target: > Several TAs in set would share the target > What if they had NOT the same target at TA level but all TAs in the > set had roughly the same target where some targets were sub-categories > of a broader target and we wnated to show this by putting the less > specific target in the 'shared' section but making some TA targets more > specific by including the more specific target in the TA within the set? > All this could be consistent, I think, with a shared target which composes > (by conjunction) with targets of any TAs at individual TA level. If a TA > target is 'A' (widget) at set level and some have a more specific 'A.a' at > TA level (medium-sized widget) then it seems that in general there is > composition with conjunction since in general 'A.a' = 'A' AND 'A.a' > > variable: makes sense if it always overrides (no sense having >1 value) > > prerequisite: always composes but maybe a disjunction is sometimes needed > - is there another way to do that? Disjunction seems to be a change to semantics > of a ta set though so I'd just as soon make it a default rule that it > is a conjunction > > predicate: makes sense not to override. Usually expect it is overridden or > there is no TA predicate when there is a TA set with shared predicate. Maybe > composition with conjugate OK as a general rules > > prescription: expect might want to be able to use an override if there is > a use case where a profile TA set overrides, say, optional TAs with > mandatory prescription. This all seems a special case so default could > be the a shared prescription is overridden by a TA one because it would > not in the normal use of a TA set/shared to change the prescription of > a contained TA. > > > tag: makes sense that it usually composes with conjunction > > norm source: as for target > > comment: as for tag > > interpretation: as for comment/norm source > > > target: compose > var: overrides > prerequisite: compose > predicate: compose > prescription: overridden? > tag: compose > norm source: compose > comment: compose > interpretation: compose > > > compose: target, prerequisite, predicate, tag, norm source, comment, > interpretation > overrides: var > overridden: prescription > > So it seems feasible that each part has a typical defaul control > of co-occurrance because each has a typical expection when it > is shared. > > This makes me suggest we could maybe just use 'shared' with > each TA Part as a direct child and the defaults could be made rules > to determine default behaviour in case of co-occurrance/concurrency. > > Then there remains a question of what to do about allowing non-default > concurrency rules but one way to do this would be to defer this to > profiles and leave the 'shared' element with an 'xs:any' for extensibility. > > So I would propose removing 'composites' and 'overrides' as child > elements of 'shared' and add rules to the spec with the proviso that > these can be supplemented or overridden in a conformance profile: > They are default rules. > > > > Best regards > --- > Stephen D Green >
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]