RE: First shot at custom tokenfilter

> -----Original Message-----
> From: Andi Vajda [mailto:vajdaosafoundation.org] 
> Sent: Wednesday, March 28, 2007 7:48 PM
> To: Ofer Nave
> 
> On Wed, 28 Mar 2007, Ofer Nave wrote:
> > I've unpacked a fresh source dist (2.1.0-1), and
copied 
> what I think 
> > is the Makefile the admin used.  Diff to the stock
Makefile:
> 
> Yeah, that failure is because you're using python 2.3
and 
> that version didn't have a set object. I warmly
recommend 
> upgrading at least to version 2.4.3.

Trying.  The admins are under heavy load right now, and the
first attempt
apparently didn't go smoothly.  I'm trying to keep that fire
burning as hot
as possible.

> That being the only failure, your PyLucene build feels
sane.

Good to know.

-ofer

_______________________________________________
pylucene-dev mailing list
pylucene-devosafoundation.org
http://lists.osafoundation.org/mailman/listinfo/pylu
cene-dev

> -----Original Message-----
> From: pylucene-dev-bouncesosafoundation.org 
> [mailto:pylucene-dev-bouncesosafoundation.org] On
Behalf Of Ofer Nave
> Sent: Wednesday, March 28, 2007 7:53 PM
> 
> > -----Original Message-----
> > From: Andi Vajda [mailto:vajdaosafoundation.org]
> > Sent: Wednesday, March 28, 2007 7:48 PM
> > To: Ofer Nave
> > 
> > On Wed, 28 Mar 2007, Ofer Nave wrote:
> > > I've unpacked a fresh source dist (2.1.0-1),
and copied
> > what I think
> > > is the Makefile the admin used.  Diff to the
stock Makefile:
> > 
> > Yeah, that failure is because you're using python
2.3 and 
> that version 
> > didn't have a set object. I warmly recommend
upgrading at least to 
> > version 2.4.3.
> 
> Trying.  The admins are under heavy load right now, and
the 
> first attempt apparently didn't go smoothly.  I'm
trying to 
> keep that fire burning as hot as possible.

Tired of waiting.  Decided to try it myself.

I grabbed the newest Python 2.5 source package.  Logged in
as regular user
'ofer', I ran ./configure --prefix=/home/ofer/tmp, make, and
make install.
Everything worked.  (I ran make test - one test failed. 
Don't know enough
to investigate, so I optimisticly ignored it.)

I then (also as user 'ofer') grabbed the PyLucene 2.1.0-1
source package,
edited the makefile to point to my new python 2.5 install in
/home/ofer/tmp,
ran make, followed by make install.  That worked, too.

I then ran my test case script using the new python
(~/tmp/bin/python
crash.py).  It doesn't seem to be crashing anymore.

Occam's razor would point to the answer being the python
threading bug fix
in 2.3.5 that someone mentioned here earlier (I have been
using 2.3.4).  No
idea if that's true, but what's important is that using 2.5
seems to make
the problem not exist, and that's good enough for me for now
(though not
good enough for users who don't have the option to
upgrade).

Sorry if I wasted your time for what turns out to be a
non-PyLucene problem.
:(

Now I need to get the admins to properly upgrade the system
python.  

-ofer

_______________________________________________
pylucene-dev mailing list
pylucene-devosafoundation.org
http://lists.osafoundation.org/mailman/listinfo/pylu
cene-dev