CADiZ

Reference manual / Further work


  1. Known bugs should always be top priority. Please make us know about them!
  2. The documentation would benefit from more examples, and from tutorials aimed at particular uses of CADi\num.
  3. The ability to decide the side-conditions of rewrite rules needs to be improved, and then effective sections of rewrite rules should be developed.
  4. Type-constrained generics could be exploited, allowing definitions of schema operations to be moved from core to toolkit.
  5. More inference rules for schema operations could be derived, possibly for the type-constrained generic forms.
  6. The dialogue boxes are too crude: there is no cut/copy/paste functionality, and forms (if we had them) would provide freedom over the order in which to answer the questions.
  7. Is there a widely-used graphical user interface that already provides editable text frames where the text can be Unicode? Is its source code available to create a variant of the text frame in which selection is of (nested) well-formed formulae rather than characters? If we were to reuse such a GUI, could the CADi\num commands be organised in pull-down menus? Would troff and the previewer still be needed for the proof tree?
  8. In proof trees, the widest rows could be largely avoided by laying out multiple leaves having the same parent vertically.
  9. Could the script manager be made to track dependencies on specific tactics and lemmas?
  10. Proof scripts can be recorded of parts of proofs. Could they be recorded in more general forms, so that they could be applied to other similar parts of the proof?
  11. Could anything be done to ease the repair of a broken proof script? For example, generalisation to make it more widely applicable, or mutation of the bit that fails followed by resumption of the remainder (a debugger for tactics).
  12. Could interchange of specifications with other tools be eased, e.g. using an XML representation?
  13. Many of the inference rules avoid seemingly necessary side-conditions by automatic renaming. Could this approach be applied to more of the inference rules?
  14. Could the functionality that displays dependency graphs be adapted to show the parents relation between sections?
  15. Can't you make it run on Windows?

IT 20-Nov-2001