(Most of) v0.94 of the Language Design Summary available

[pdmosses@brics.dk]

I have just installed what is now ready for v0.94: basically
everything *except* Part III on architectural specifications.  
I *hope* to finish that bit later today; I'll announce it on this 
list as soon as it is installed.

The URL for v0.94 is as before:

  http://www.brics.dk/Projects/CoFI/DesignProposals/Summary/

and the formatted versions for printing are at URL:

  ftp://ftp.brics.dk/pub/BRICS/Projects/CoFI/DesignProposals/Summary.ps.Z
  ftp://ftp.brics.dk/pub/BRICS/Projects/CoFI/DesignProposals/Summary.dvi

Anyway, the basis for the discussion concerning Part III is already
available: the study note [PDM-2], Bernd's response to it today on
this list, and the previous papers and study notes on the topic.  
(But I'm finding it not so easy to summarize all that for v0.94...)


CHANGES IN V0.94  

The main changes in version 0.94 are listed below. Significant changes to the
abstract syntax are indicated by change-bars in the complete listing of the
Abstract Syntax in the appendix.

  Name
    The current favourite proposal for the name of `COFI Common Language for
    Algebraic Specification' is: CASL.

  Basic Specifications
    Explicit qualification of function and predicate symbols by profiles is
    used to ensure unambiguity of terms in the underlying institution. The
    notions of well-sortedness and expansions of atomic formulae are used to
    define well-formedness.

  Basic Concepts
    The signature morphism component has been removed from constraints.

  Basic Constructs
    Compound identifiers have been moved to Part II.

    Global variable declarations are to be discussed.

  Subsorted Concepts
    Weak monotonicity has been removed.

  Subsorted Constructs
    Subsort declarations are now separate from subsort declarations.

  Structuring Concepts
    This section has been completely rewritten.

  Structuring Constructs
    The section has been completely reorganized.

    Cyclic sections of libraries have been removed.

    Generic (parametrized) specifications are described in more detail.

    Mappings are now between possibly-qualified symbols.

    Compound sorts are now monotonic wrt. subsort embeddings.

  Architectural Specifications
    This part is still being changed...

My assessment of which main issues are still being discussed is as
follows: 


MAIN ISSUES FOR DISCUSSION

  1. Is the proposed notion of compound identifiers for symbols
     satisfactory (for use in connection with instantiation of parametrized
     specifications, at least)?
  2. Are the constructs for generic (parametrized) specifications
     satisfactory? How strict should the well-formedness conditions be
     regarding declaration and use of parameters? Should it be allowed to
     fix parts of parameters?
  3. Are the constructs for architectural specifications satisfactory?
     Should only one way of expressing compositions be allowed, or both? Is
     the assumption of unique implementation reasonable? Are generic
     architectural specifications needed? Are specifications parametrized
     by algebras to be included?
  4. Are local definitions of specifications needed in structured
     specifications?


----   --------------------------------------------
\  /  | Peter D Mosses         <pdmosses@brics.dk> |
CoFI  | Common Framework Initiative  - Coordinator |
/  \  | WWW URL: http://www.brics.dk/Projects/CoFI |
----   --------------------------------------------