<div dir="ltr">Is there anything to learn about this bug that could lead to avoiding such bugs in the future? Perhaps some simple property that was being violated that one could write a checker for (maybe running Racket in some debug mode)?<div>
<br></div><div>Robby</div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Fri, Apr 19, 2013 at 11:20 AM, Matthew Flatt <span dir="ltr"><<a href="mailto:mflatt@cs.utah.edu" target="_blank">mflatt@cs.utah.edu</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">At Thu, 18 Apr 2013 22:23:37 -0600, Doug Williams wrote:<br>
> I did have DrRacket crash twice on startup. I'm running the 64-bit version<br>
> under Windows 8 64-bit.<br>
<br>
The stars have aligned this morning!<br>
<br>
We've had reports of crashes specific to 64-bit Windows, but I had not<br>
been able to replicate the crash. This morning, it turns out that I'm<br>
able to provoke a crash easily and consistently (by trying the same<br>
things that I've always been trying).<br>
<br>
That crash that I see in the stack-unwinding code for generating JIT<br>
stack traces, which is Win64-specific and added September 2012 (after<br>
v5.3, before v5.3.1). That functionality is particularly likely to be<br>
triggered when you Run or Check Syntax a program. The bugs could easily<br>
manifest as either an immediate crash or a freeze. And the bugs could<br>
easily have no effect, depending on Racket thread scheduling and the<br>
layout of the stack. So, I think this bug is consistent with all of the<br>
reports.<br>
<br>
Even better, it turns out that I spent much of yesterday staring at the<br>
stack-unwinding code (for ARM), so the details were all paged in.<br>
<br>
I've pushed a repair, and I'll ping you when a build is ready to try out.<br>
<br>
_________________________<br>
Racket Developers list:<br>
<a href="http://lists.racket-lang.org/dev" target="_blank">http://lists.racket-lang.org/dev</a><br>
</blockquote></div><br></div>