On Mon, 19 Nov 2001, Stephan Rudlof wrote:

> Bijan Parsia wrote:

[snip]
> > But so is:
> > 
> >   testPPIf: aBool
> >         aBool ifTrue:[^5]
> >                 ifFalse:[^nil]
> > 
> > (ok, perhaps only in this simple example).
> > 
> > And in that case PPing gets:
> > 
> >   testPPIf: aBool
> >         aBool
> >                 ifTrue: [^ 5]
> >                 ifFalse: [^ nil]

There are too many examples. Which is *my* fault! :)

> Fully correct!
> 
> *Wrong* would be
>     testPPIf: aBool
>           aBool
>                   ifTrue: [^ 5]
> , since this would return self.

Yes, but I didn't *want* that :) My point is that only very specific
equivalences are transformed (the ^...ifTrue:..ifFalse[nil] but not the
^...ifTrue: [^...] ifFalse: [^...] was transformed(, whereas others are
not. This is for a good reason, of course. That reason being that the
compiler doesn't *make* those other transformations. But this ties the
pretty printer to specific features of the compiler which can easily
change.

Imagine hooking up the pretty printer to the Refactoring Browser's
parsing/lint mechanism so that *pretty printing* would produce
refactorings! That might actually be *useful* for checking out what a
refactoring would do to the code, but I sure wouldn't want to call it
pretty printing :) (of the original code).

Cheers,
Bijan Parsia.