Issue CLOS-ERROR-CHECKING-ORDER Writeup

Issue:        CLOS-ERROR-CHECKING-ORDER

Reference:    dpANS 12.24

                Section 7.6.5 Keyword Arguments in Generic Functions and Methods

                Section 7.6.6 Method Selection and Combination

                no-applicable-method, p.7-47

              X3J13/92-3248 David Moon comment #48

Category:     CLARIFICATION/CHANGE

Edit History: Version 1, 12/13/92, Kim Barrett

Status:	      Proposal NO-APPLICABLE-METHOD-FIRST passed (10+1)-0 on

              letter ballot 93-302.


Problem Description:


 There is no specification of the ordering between the check for

 no-applicable-method and the check for unexpected keyword arguments.

 Specifically, section 7.6.5 states that if a generic function is passed a

 keyword argument that no applicable method accepts, an error should be

 signaled, while section 7.6.6 states that if there are no applicable methods

 then the generic function no-applicable-methods is called.


Proposal (CLOS-ERROR-CHECKING-ORDER:NO-APPLICABLE-METHOD-FIRST):


 Check for no applicable method first.  Change the end of the last paragraph

 of section 7.6.6 from


    If a \term{generic function} is called and no \term{methods} are

    \term{applicable}, the \term{generic function} \funref{no-applicable-method}

    is invoked.


 to


    If a \term{generic function} is called and no \term{methods} are

    \term{applicable}, the \term{generic function} \funref{no-applicable-method}

    is invoked, with the \term{results} from that call being used as the

    \term{results} of the call to the original \term{generic function}.  Calling

    \funref{no-applicable-method} takes precedence over checking for acceptable

    keyword arguments; \SeeSection{... section name for 7.6.5 ...}.


Editorial Impact:


 Changing or adding a couple of lines of text.


Rationale:


 It was clearly intended that the call to no-applicable-methods be for value,

 rather than for effect.  It just wasn't actually specified that way anywhere.


 Adopting the suggested order would remove an ambiguity from the specification

 and would make the no-applicable-method feature a little more useful.