Announcing SpiderMonkey 1.8 Release Candidate 1

[Jason Orendorff]

The Mozilla JavaScript team is pleased to announce SpiderMonkey 1.8
Release Candidate 1.

Release notes
https://developer.mozilla.org/en/SpiderMonkey/1.8

Source code
http://ftp.mozilla.org/pub/mozilla.org/js/js-1.8.0-rc1.tar.gz
MD5 checksum: eaad8815dcc66a717ddb87e9724d964e

SpiderMonkey 1.8 is the JavaScript engine that shipped in Firefox 3.0.
It is much faster than SpiderMonkey 1.7 and contains a few new language
features and API features, described in the release notes.

Please help us by testing this release candidate and let us know about
your experiences by posting here. Or, file bugs at
https://bugzilla.mozilla.com under Product: Core, Component: JavaScript
engine.

--
The Mozilla JavaScript Team

[Paul Barnetta]

Hi there,

> Please help us by testing this release candidate and let us know about
> your experiences by posting here.

I posted a question on the group last month ("Issues with
Multi-threading?") and have bug 478336 in bugzilla corresponding to
this. Neither have attracted any comments and I was keen to see how this
release candidate handled the test application I submitted then.

It seems that for me this release is giving the following assertion
failure when run with a number of threads:

Assertion failure: JS_PROPERTY_CACHE(cx).disabled >= 0, at jsinterp.c:509

The code works fine on 1.7 and with this release candidate *when run
with just a single thread* (or a few threads, say less than ~5 for me on
my laptop at the moment). If I run with 100 threads and run the
application 100 times I get ~3/4 failure rate:

$ for ITER in $(seq 100); do ./a.out 2>&1; done | sort | uniq -c
74 Assertion failure: JS_PROPERTY_CACHE(cx).disabled >= 0, at
jsinterp.c:509

This ratio is similar to what I posted last month, but the assertion is
different to the ones I was getting then:

$ for ITER in $(seq 100); do ./a.out 2>&1; done | sort | uniq -c
4 Assertion failure: first, at ../jscntxt.cpp:314
69 Assertion failure: rt->state == JSRTS_UP || rt->state ==
JSRTS_LAUNCHING, at ../jscntxt.cpp:465

I was just wondering whether anyone could provide any advice. Is there
an issue with threading in 1.8, or an issue with the code? I'd welcome
any comments!

Regards,
Paul

[Jason Orendorff]

On 3/9/09 4:51 PM, Paul Barnetta wrote:
> I posted a question on the group last month ("Issues with
> Multi-threading?") and have bug 478336 in bugzilla corresponding to
> this. Neither have attracted any comments and I was keen to see how this
> release candidate handled the test application I submitted then.

I've commented in the bug. We currently don't have a fix for this
issue, as far as I know. It is a significant regression, and I'll track
it to see what we can do before 1.8 final.

-j

[Paul Barnetta]

> I've commented in the bug. We currently don't have a fix for this
> issue, as far as I know. It is a significant regression, and I'll track
> it to see what we can do before 1.8 final.
>
> -j

Thanks Jason; it's just reassuring to know someone else can see the
issue too :) Will look forward to testing the release more later..

Regards,
Paul