[BC] Block Closures, Version 2

Scott A Crosby crosby at qwes.math.cmu.edu
Mon Feb 18 20:29:31 UTC 2002


On Mon, 18 Feb 2002, Anthony Hannan wrote:

> John M McIntosh <johnmci at smalltalkconsulting.com> wrote:
> > Anthony Hannan wrote:
> > >In the bug you described it looks like fp is off by two words because
> > >longAt(fp-4) = a MethodContext2 instead of a CompiledMethod2.
> >
> > So do you have a fix?
>
> Unfortunately I don't.  I don't get this bug on my machine and I don't
> see any problems when looking at the code.  I'm baffled.  How can the
> same code work on my machine but on no one elses (mac, linux, or
> windows)?  I downloaded everything from scratch from Squeak,
> sourceforge, and my BC page as if I was a new user and it still works.

I'd guess that its because the generated VM/image differ based on
something that is outside of the raw code.

Is there some patch version inconsistency between the page and what you
actually use? Patch application order requirement? Was some change put in
the wrong changeset (say, a VM fixup placed in the image-translation
changeset, or vice versa. So you must load all changes before translating
an image or building a VM.)

Or, for example, do things break if you do:
   BCSystemTracer new writeImageFile: 'yourBC.image'
INSTEAD OF
   Time millisecondsToRun:
          [BCSystemTracer new writeImageFile: 'yourBC.image']

When I've been working on it, I've created two throwaway clean images,
load the VM changes into one, and the translator into the other. Do you do
things differently?

Scott





More information about the Squeak-dev mailing list