Dear Dave,:
> Suppose I file in a method:
>
> foo ^MissingClass open
>
>
> MissingClass becomes an entry in Undeclared. Why shouldn't filing in
> this method also create an entry in Undeclared?:
>
> bar ^Smalltalk.AnotherMissingClass open
It is a known (and undesirable) feature of the Undeclared search that it
does not reveal dot-referenced Undeclared (ARs 49017 and 50286).
We should of course mention this in the Known Limitations release-notes
section on Undeclared and offer solutions - I'll address that.
We should also address the bug (feature, I could say - this touches on
basic questions about the original intent of the Uncdeclared cache - but
if so its a feature I want to fix/update).
(Obviously, in your example, the code should eliminate the pointless
dot-reference. In general, good structuring with general and specific
imports, and exploiting the package #namespace property, to eliminate
dot-references in methods, is good practice. However there will always
be exceptions, especially perhaps in code that is in early development.)
Yours faithfully
Niall Ross
_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc