Node::reverse*() cleanup

[porten at froglogic.com]

On Sun, 2 Nov 2003, Darin Adler wrote:

> > +    CaseClauseNode(Node *e, StatListNode *l)
> > +      : expr(e) { if (l) { list = l; l->list = 0; } else { list = 0;
> > } }
>
> The above code looks wrong to me. I think it should be list = l->list,
> not list = l.

True.

> This indicates that there aren't sufficient tests for

Or someone (me) forgot to run one of them after a last minute change.
khtmltests/js/statements.js would have shown this bug. But I've now also
added a test that tests this case more explicitly.

> I applied the "leave out the parameter to separate out the 0 case"
> optimization to the above CaseClauseNode constructor, one place where
> you did not do it.

Did so as well now. Would have prevented the error I guess.

> The patch didn't include ObjectLiteralNode or ArgumentListNode, but I
> guess that's because you did that in an earlier patch. I was slightly
> surprised by that. I had expected to see it all at once; I didn't
> realize you had already landed part of this.

Yes. Mentioned it shortly in a previous mail. The first patch was applied
in February.

> Here's the patch I have so far; I figured you might want to see my
> version.

Applied your additions.

Thanks for the review,

Harri.