Re: [Chicken-hackers] [PATCH] Fix #1620 by ignoring captured state of replaced variables

[Peter Bex]

On Sun, Jun 30, 2019 at 08:55:09PM +0200, address@hidden wrote:
> Yes, this appears to be the correct way - I wondered why there was not
> an additional constraint on assignment, but since the order of optimizations
> is not easily seen in advance, one has to be careful about many assumptions.

Yeah, these things aren't easy.

> I think your reasing is correct, but must ask everybody to let me review this 
> for
> a certain time before we commit this patch.

Thank you for taking the time!  I was hoping you would take a look at
it :)

Note that I forgot to mention in my original mail that I also removed
the constraint that the variable should only be referenced once.  As far
as I can tell, if one instance of a non-assigned variable can be replaced
by the variable it aliases, all other instances should be replaceable in
the same way.

And the first hunk of the patch is not a change in functionality.  It
just changes the comment to reflect reality, but you probably noticed
that already.

Cheers,
Peter

signature.asc
Description: PGP signature