Mailing-List: contact dev-help@harmony.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@harmony.apache.org
Received-SPF: neutral (herse.apache.org: 195.212.29.138 is neither permitted
 nor denied by domain of mark.hindess@googlemail.com)
Message-Id: <200705180809.l4I895Ev016072@d06av04.portsmouth.uk.ibm.com>
In-reply-to: <464AD88E.7000102@googlemail.com> 
References: <200705160820.l4G8KlAU014458@d12av04.megacenter.de.ibm.com>
 <f944d83d0705160257m2f7ebcev457192c6bbe8b702@mail.gmail.com>
 <464AD88E.7000102@googlemail.com>
Comments: In-reply-to Oliver Deakin <oliver.deakin@googlemail.com>
   message dated "Wed, 16 May 2007 11:10:22 +0100."
From: Mark Hindess <mark.hindess@googlemail.com>
To: dev@harmony.apache.org
Subject: Re: [classlib] hy.no.sig option 
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Date: Fri, 18 May 2007 09:09:03 +0100


On 16 May 2007 at 11:10, Oliver Deakin <oliver.deakin@googlemail.com> wrote:
> +1
> 
> Agreed. I dont think we lose anything by turning off the signal handler 
> as default. I have tested this mode and have not seen any problems with 
> it so far.

Actually we do lose something... drlvm stops working.  It looks like 
something must be using jsig_primary_sigaction which is removed from
classlib but which must be present in the IBM VME.

-Mark.

> Alexey Varlamov wrote:
> > +1. AFAIU this is also a step towards resolving the glorified hythr
> > libraries conflict.
> >
> > 2007/5/16, Mark Hindess <mark.hindess@googlemail.com>:
> >>
> >> For a while now, I've been building classlib with the hy.no.sig option
> >> to avoid the problems of both classlib and the VM trying to install
> >> master signal handlers.  Is it time we made this the default?  Has
> >> anyone else been testing this?
> >>
> >> FYI: the easiest way to test it today is just doing:
> >>
> >>  echo hy.no.sig=true >>$HOME/.harmony-classlib.properties
> >>
> >> Of course, it really shouldn't make any practical difference since the
> >> old classlib signal handler was only invoked in the event of a crash.
> >>
> >> I propose we change the default from false to true.  Does this seem
> >> reasonable?  This will mean that no handler will be registered until the
> >> VM is initialized but I don't think this is a significant issue - since
> >> crashes before the VM is initialized should (hopefully) be pretty rare.
> >>
> >> Ultimately, I'd like to remove the option altogether since maintaining
> >> duplicate signal handlers doesn't offer any significant benefits.
> >>
> >> Regards,
> >>  Mark.
> >>
> >>
> >>
> >
> 
> -- 
> Oliver Deakin
> Unless stated otherwise above:
> IBM United Kingdom Limited - Registered in England and Wales with number 7415
> 98. 
> Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU
>