make more systematic transformation between XML (xtc, cpf) and data

[jwaldmann]

current situation is:

• abstract model for is data declarations
  • https://github.com/jwaldmann/haskell-tpdb/blob/master/src/TPDB/Data.hs (benchmarks) ,
  • https://github.com/jwaldmann/haskell-tpdb/blob/master/src/TPDB/CPF/Proof/Type.hs (proofs)
• concrete realization is XML schema
  • http://cl2-informatik.uibk.ac.at/mercurial.cgi/TPDB/file/912be42cb054/xml/xtc.xsd (benchmarks),
  • http://cl2-informatik.uibk.ac.at/mercurial.cgi/CPF/file/f5ddfdf4ff7f/cpf.xsd (proofs)

Translation is ad-hoc - that is, not uniform (generic) over the data type declarations. (For reference, here is a generic transformation: http://repetae.net/repos/DrIFT/src/Rules/Xml.hs)

So my current code is a mess of

• ad-hoc transformations, cf. https://github.com/jwaldmann/haskell-tpdb/blob/master/src/TPDB/CPF/Proof/Read.hs
• parsing and printing are officially unrelated (and sometimes one is missing)

So, we have a problem similar to mapping data declaration <-> existing data base schema. In persistent , you can specify details of the mapping via annotations to the model. This needs template haskell, see https://github.com/yesodweb/persistent/blob/master/persistent-template/Database/Persist/TH.hs

Plan (long-term)

1. describe a simple and fully generic data <-> XML mapping.
   • this should map an object C1 { a1 = foo } of type data T = C1 { a1 :: A1, .. } | .. to <T><C1><a1> ...</a1>..</C1></T>, so that is has all information
2. describe a set of ("ad-hoc") modifications that can be applied to give the desired result. E.g.,
   • use different name for element than for constructor
   • drop some of these elements (T,C1,a1)
   • describe mapping for primitive types (to attribute, or to contents)
   • describe handling of implicit collections (list)