ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Steve Loughran" <stev...@iseran.com>
Subject Re: ExecWatchdog test failing
Date Sat, 02 Mar 2002 00:26:13 GMT

----- Original Message -----
From: "Stephane Bailliez" <sbailliez@apache.org>
To: "Ant Developers List" <ant-dev@jakarta.apache.org>
Sent: Friday, March 01, 2002 16:00
Subject: Re: ExecWatchdog test failing


> ----- Original Message -----
> From: "Steve Loughran" <steve_l@iseran.com>
> [...]
> > that is why I have one. Only cost $120 to stick a new PIII beside the
old
> > one, helps me test things like winXP for thread safetly.
>
> > Not a watchdog bug on an MP system, just a timing clock resolution
thingy
> in
> > the test itself
>
> btw why '200ms' ?
>
> Even though they are supposed to run on different cpu I would expect the
> startup and shutdown of the VM to be a little bit more than 0ms.... I
don't
> think that the memory model on the sun vm is this good for thread mapping
> and would expect things to be far more unpleasant on the JRockit vm which
is
> designed to take better advantage of MP boxes because of its MxN thread
> mapping. It can even do parallell garbage collection. (experimental as of
> today). The startup though is a little bit slower since it compiles all
code
> at startup.


I chose 200mS because of timer resolution...sometimes a sleep() doesnt last
as long as it says. Also on an MP system the time to shutdown the JVM may be
incurred after the other app has got its finished signal. Something like
50mS would be less pessimistic.


--
To unsubscribe, e-mail:   <mailto:ant-dev-unsubscribe@jakarta.apache.org>
For additional commands, e-mail: <mailto:ant-dev-help@jakarta.apache.org>


Mime
View raw message