harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Weldon Washburn" <weldon...@gmail.com>
Subject Re: Re: [drlvm][threading] is harmony-1951 a bug or a feature?
Date Wed, 22 Nov 2006 04:38:52 GMT
On 11/21/06, Rana Dasgupta <rdasgupt@gmail.com> wrote:
>
> On 11/21/06, Weldon Washburn <weldonwjw@gmail.com> wrote:
> >
> > On 11/21/06, Elford, Chris L <chris.l.elford@intel.com> wrote:
> > >>In general, I agree that spurious wakeups are something to be avoided
> by
> > >> the implementation where possible [personally, I'd like to see them
> > >> disallowed].  However, given that the spec allows spurious wakeups,
> it
> > >> could actually be a valuable development tool to have an option to
> > >> insert them extensively to help developers identify places where they
> > >> have inadequately protected themselves against true spurious wakeups.
> > >> Therefore, I could see this partially as a feature.
> >
> >
> > >Yes!  I like this idea.  It should not be too hard to add to drlvm when
> > the
> > >time is right.
> >
> > This makes no sense to me at all, sorry. There are possibly other ways
> to
> > develop good coding advisors. Making the VM a booby trap for all kinds
> of
> > corner cases is not one of them :-)


For a production JVM, you are right.  We don't want innocent users to set
off booby traps by playing around with the command line.  But this feature
might still make sense for an engineering debug build.





-- 
Weldon Washburn
Intel Enterprise Solutions Software Division

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message