Yes, those two tar balls I uploaded to your mirandabanda last week,
from our other thread, "Reproducible problem with String>>#<=".  One
of the tar balls has the image saved with the debugger open on what I
think looks like the invalid stack frame; where you can see the VM has
chosen to execute the code in the wrong order.  (PS -- I don't think
it has to do with becoming on the stack, but I'm not positive about
that).

(PPS -- in that image, note that another debuger will pop up from
startup code when the image launches about a file not found, just
ignore that one).

The other tar ball is one that can reproduce the problem;
unfortunately, not quickly and not always in the same place, but it
did eventually happen when I tried to load several thousand mcz's.
Unlike my past tests, this job is a single-core task, not multicore,
so a bit easier on your machine in case you decide to run it.

But I didn't try it iwth the latest Spur 3312 VM yet.  However, I have
the 3312 coglinuxht VM doing it on box4 right now for 13000 mcz's.  It
is almost half done, other than that problem yesterday ("halt: no free
methods; cannot compact.") it has been running smoothly.  Only about
halfway done now probably another couple of days to finish..


On Tue, Apr 21, 2015 at 9:09 PM, Eliot Miranda <eliot.miranda at gmail.com> wrote:
>
>
> On Tue, Apr 21, 2015 at 9:14 AM, Chris Muller <ma.chris.m at gmail.com> wrote:
>>
>> Hi Eliot, I am working on restoring the history function right now.
>> I'm currently running a job which is rebuilding it on box4, but box4
>> is excruciatingly slow and I'm just using Cog instead of Spur due to
>> that String <= problem (the invalid stack frame).
>
>
> Hmm, we should fix that.  Does it still happen with 3312?  Got a
> reproducible case if so?
>
>>
>> I hope to have it working again later this week.  Sorry for the
>> inconvenience!
>>
>> On Tue, Apr 21, 2015 at 11:04 AM, Eliot Miranda <eliot.miranda at gmail.com>
>> wrote:
>> > Hi All,
>> >
>> >     looks to me that classCommentBlank should /not/ check for an
>> > existing
>> > comment but simply answer the computed template.  Agreed?  I thought I'd
>> > check the mc history to be more informed, but I can't connect to
>> > box4.squeak.org.  Where is the history these days?
>> >
>> > --
>> > best,
>> > Eliot
>
>
>
>
> --
> best,
> Eliot