Jack Schwartz wrote: > Ethan Quach wrote: >> >> >> Jack Schwartz wrote: >>> Karen Tung wrote: >>>> Jack Schwartz wrote: >>>>> >>>>> General: >>>>> >>>>> In general, the parser places the requirement on the derived >>>>> profiles project that the format of the tags be in an easily >>>>> searchable format, for example that they contain a common prefix >>>>> which can be searched for. >>>>> >>>>> >>>> I don't see why the parser would place requirement on the derived >>>> profile project on the format >>>> of the tags or whatever. Since the new parser design doesn't >>>> automatically do syntactic validation, >>>> it will be up to the derived manifest client program or AI setup >>>> that accepts derived manifest >>>> to perform the derivation whenever it is appropriate, >>>> and call the parser's function to do syntactic validation whenever >>>> it feels it is appropriate. >>>> From the parser's point of view, I don't think it should even need >>>> to know about whether it is >>>> a derived manifest or the regular manifest. >>> It is true that the parser shouldn't care about the data it handles. >>> >>> This requirement is more of a restriction on the derived profiles >>> project to structure its data so that the parser can be used to >>> search it easily. >> >> Seems like we should declare this the other way >> then. If derived manifests is going to need something >> fancy, its going to have the requirement on the parser >> component to provide for that fanciness. > I suppose you are right. However, you should be able to do what you > need to do with the parser we are currently planning on providing, if > your data is set up properly. > > Thanks, > Jack > > P.S. Actually, earlier I said it was the names of tags need to be > structured a certain way; really it's the values of those tags. Then > the parser can search for values of a certain format and derived > manifest code can process only those fields. I would think that having a well defined tag shouldn't be a parser's requirement on derived manifest project. The derived manifest project also shouldn't depend on the parser for providing any fancy features.
The parser will provide for searching on elements and values based on xpaths. The derived manifest project will look at this parser feature and decide during IMPLEMENTATION TIME how the data should be structured so that it is easily searchable and updatable...etc.. So, there's really no dependency between the 2 projects, I think. --Karen
