lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jamie M <jami...@yahoo.com>
Subject Re: New FieldSortedHitQueue uses Java 1.4 feature
Date Tue, 09 Mar 2004 14:37:14 GMT

--- Erik Hatcher <erik@ehatchersolutions.com> wrote:
> On Mar 8, 2004, at 12:54 PM, Doug Cutting wrote:
> > Jamie M wrote:
> >> And I've come up with two workarounds for this
> problem
> >> under an ibm 1.3.1 jvm:
> >> 1) disable jit compliation for this method by
> setting
> >> this environment variable:
> >>
>
JITC_COMPILEOPT=SKIP{org/apache/lucene/store/OutputStream}{writeInt}
> >
> > Good job finding this!
> >
> > Is it possible to add this to the Ant build
> script?
> 
> I'm not sure if this can be made part of the Ant
> script or not.  If it 
> has to be a true environment variable, I don't think
> it can.  But if it 
> can be made to be part of the compiler arguments
> somehow, then yes.
> 
> Jamie - if you can see if this works by tweaking the
> Ant build file, we 
> could certainly make it part of the the build
> process in some manner.
> 
> 	Erik

Unfortunately handling of this jvm bug can't be done
in the ant build script.  If someone wants to use
Lucene 1.4 on an ibm 1.3.1 jdk they will have to set
the environment variable before starting the jvm.  Or,
if this is inconvenient (as with my case, e.g. when
deploying a web application to a number of servers
that the developer doesn't have access to, or
permission to make environment changes to) then a
custom build of Lucene can be done to explicitly
linline the methods the JIT is messing up.

The easiest way for 1.3.1 users to not be affected by
this problem would be to commit my inlining patch,
however the code that the JITC began having problems
with about 3 months ago hasn't changed for over a year
according to cvs.  This suggests that code changes to
some of lucene's classes can surface JITC bugs
relating to other (unchanged) classes, which may make
coding around such JITC bugs a moving target.

I've opened a PMR (problem management record) with
IBM's Java Support for this problem - the number is:
07456,001,866

As the status of this PMR changes I'll post details to
this thread.

jamie


__________________________________
Do you Yahoo!?
Yahoo! Search - Find what you’re looking for faster
http://search.yahoo.com

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


Mime
View raw message