IBM JIT 1.3.1 vs. Lucene?

[Alex Chaffee <al...@jguru.com>]

On the jGuru Lucene FAQ, I discovered the following:

http://www.jguru.com/faq/view.jsp?EID=1152238

"Apparently there is a bug in IBM's JIT code in JDK 1.3.1. To work around
it, disable JIT for the org.apache.lucene.store.OutputStream.writeInt method
by setting the following environment variable:
JITC_COMPILEOPT=SKIP{org/apache/lucene/store/OutputStream}{writeInt} "

However, I am now using Lucene on WebSphere using IBM's JDK 1.3.1,
apparently successfully. How do I know if I'm vulnerable to this bug? What
are the cases in which it could manifest? Could IBM have fixed the bug in an
update?

The reason I don't just set the environment variable is that it's
politically difficult to change any configuration of the servers at my
client. We have a different workaround -- a homegrown search index using
hashtables -- but we'd prefer to use Lucene... but not if there's a chance
of getting a bug in production. We're down to the wire and need to decide
whether to use Lucene or not for this deployment... Can you shed any light
on this?

Thanks much -

 - Alex Chaffee
alex@jguru.com
http://www.purpletech.com/




---------------------------------------------------------------------
To unsubscribe, e-mail: lucene-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: lucene-user-help@jakarta.apache.org

Re: IBM JIT 1.3.1 vs. Lucene?

[Otis Gospodnetic <ot...@yahoo.com>]

I do not recall anyone fixing this issue (there was/is a case open in
Bugzilla).  2 AM, time for bed.

Otis

--- Alex Chaffee <al...@jguru.com> wrote:
> On the jGuru Lucene FAQ, I discovered the following:
> 
> http://www.jguru.com/faq/view.jsp?EID=1152238
> 
> "Apparently there is a bug in IBM's JIT code in JDK 1.3.1. To work
> around
> it, disable JIT for the org.apache.lucene.store.OutputStream.writeInt
> method
> by setting the following environment variable:
> JITC_COMPILEOPT=SKIP{org/apache/lucene/store/OutputStream}{writeInt}
> "
> 
> However, I am now using Lucene on WebSphere using IBM's JDK 1.3.1,
> apparently successfully. How do I know if I'm vulnerable to this bug?
> What
> are the cases in which it could manifest? Could IBM have fixed the
> bug in an
> update?
> 
> The reason I don't just set the environment variable is that it's
> politically difficult to change any configuration of the servers at
> my
> client. We have a different workaround -- a homegrown search index
> using
> hashtables -- but we'd prefer to use Lucene... but not if there's a
> chance
> of getting a bug in production. We're down to the wire and need to
> decide
> whether to use Lucene or not for this deployment... Can you shed any
> light
> on this?
> 
> Thanks much -
> 
>  - Alex Chaffee
> alex@jguru.com
> http://www.purpletech.com/
> 
> 
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: lucene-user-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: lucene-user-help@jakarta.apache.org
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: lucene-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: lucene-user-help@jakarta.apache.org