There is a problem with all these schemes that use code as signs, which
is that as far as automated analysis goes, this makes SystemDictionary
dependent on all those classes, and preserves the mess we're trying to
get rid of in the first place.

This is not a completely theoretical problem, because there is a tool
that actually does this analysis, MudPie.

But I don't see what the problem is with make SystemDictionary MNU bring
up an "implementor of". Or mabye make it an option in the error
notifier, whenever there's an MNU, whatever the class.

Daniel

Bob Arning <arning at charm.net> wrote:
> On Fri, 13 Jun 2003 11:51:20 +1200 (NZST) "Richard A. O'Keefe" <ok at cs.otago.ac.nz> wrote:
> >    "Smalltalk condenseChanges
> >     is now (SourceManager condenseChanges).
> >     Smalltalk sourcesName
> >     is now (SourceManager sourcesName).
> >     Smalltalk bytesLeft
> >     is now (MemoryManager bytesLeft).
> >     ..."
> >
> >I want this stuff where the browser can find it.
> >For example, suppose I type
> >    Smalltalk bytesLeft
> >in a workspace, and get a MNU.  Then I can ask for "sources with it",
> >and be pointed right to the place I want.
> 
> Well, finding senders is quite a bit faster than finding method source containing something, so why not write
> 
> 	[Smalltalk condenseChanges] isNow: [SourceManager condenseChanges].
> 	[Smalltalk sourcesName] isNow: [SourceManager sourcesName].
> 	[Smalltalk bytesLeft] isNow: [MemoryManager bytesLeft].
> 
> and you can search for the old version and find the new (or search for the new and find the old). Plus, you can search for senders of #isNow: to find all such changes.
> 
> Cheers,
> Bob