Re: nuxeo.lucene query

Julien Anguenot <[email protected]>
Newsgroups gmane.comp.web.zope.cps.devel
Organization Nuxeo SAS
Message-ID <[email protected]>
Hi David,

David Pratt wrote:
> Hi Julien. Thanks. I'll look forward to receiving the credentials.

It should be in your mailbox, I just sent them.

> I've not used svn:externals so perhaps I'll create the package first and
> if I am uncertain about getting the externals property right I'll ask
> for help on this.

right. Do the changes in a branch at first and we will check this out after.

> I am trying to build a more recent gcc4.1 and gcc4.2 tonight and see if
> 1.9.1 and 2.0.0 will pass the tests since Andi has indicated his success
> on the list. I am on FreeBSD 6 at the moment.

right. Keep me posted because It doesn't work for me on a FC5.

> I have not used or tried NXLucene yet but have been aware of the nuxeo
> project. I been experimenting with lucene for a while ago so am curious
> about the performance with xmlrpc - whether it is good or too slow. What
> is your experience thus far?

Performances are a real win as expected. The only problem is that using
gcj is a pain to setup in production because, for instance, PyLucene
1.9.1 only works with gcc 3.4.6...

> I have a couple of indexes I want to test. I guess the other factor is
> resources. How beefy is the server (RAM) you are running and how large
> are your indexes. 

2go RAM and indexes over 2Go for the moment on one of our biggest
project using NXLucene at Nuxeo.

> I see in the docs that you are using the older gcc.
> This creates some limitations in dealing with indexes beyond a couple of
> GB, I believe. I have not yet made any indexes of this magnitude at this
> point, but would like to understand any issues or limitations. I am not
> on your mailing list but will subscribe shortly. Many thanks.
>

We have this issue. That's why I implemented a Python directory to avoid
this limit :

http://svn.nuxeo.org/trac/pub/browser/NXLucene/trunk/src/nxlucene/store

But this is slower...

Check the PyLucene-dev mailing list arvhices of June for more information.

Cheers,

	J.

> Julien Anguenot wrote:
>> Hi David,
>>
>> David Pratt wrote:
>>> Hi Julian. Thanks for your replies. I can split the rss into its own
>>> package but I will need a svn account. 
>>
>> Great. I'll send you the credentials during the week-end.
>>
>>> I'll make the separate package
>>> nuxeo.rsslucene and make license ZPL so it can work for both. I'll
>>> modify the trunk of nuxeo.lucene to import the new package instead of
>>> nxlucene.rss. I can look for any dependencies in NXLucene for the rss
>>> package and modifiy these on the trunk if you would also like this.
>>
>> right sounds good. I think I would use svn:externals myself.
>>
>>> I'll fix the typos also in nuxeo.lucene while I am at it. 
>>
>> right.
>>
>>> Are you
>>> concerned about updating for pyLucene 2.0.x. pyLucene 2.0 has been
>>> around for a while now. It is looking more promising for builds to work
>>> on gcc 4.1.0. I had built successfully built an ealier pyLucene on
>>> FreeBSD a while back but had some trouble trying to build latest but am
>>> going to try again on a newer FreeBSD version and newer gcc. I kept my
>>> notes around from the first time. Many thanks.
>>
>> Of course. Once you got the account feel free to do a branch and migrate
>> it to PyLucene 2.0. That would be definitely great. I checked a bit.
>> some part of the Field API I was using with the 1.9.x version of Lucene
>> has been deprecated on the 2.x branch. So there is a little bit of work
>> here I guess.
>>
>> I didn't personally succeed, couple of weeks ago, to run a working
>> 4.1.0, nor the trunk compiled versions of GCC with PyLucene. But Andi
>> from the OSAF seems to succeed in doing that.
>>
>> If you need help I'll be around looking a little bit more often at the
>> list than those past weeks...
>>
>> Regards,
>>
>>     J.
>>
>> P.S : Are you using NXLucene yourself at the moment ?
>>
>>> Julien Anguenot wrote:
>>>> Hi David,
>>>>
>>>> David Pratt wrote:
>>>>> Hi. I am just checking out the nuxeo.lucene (ZPL) package but
>>>>> noticed a
>>>>> dependency on nxlucene (LGPL) for two imports in catalog.py:
>>>>>
>>>>> around line 37:
>>>>>
>>>>> from nxlucene.rss.resultset import ResultSet
>>>>> from nxlucene.rss.adapter import PythonResultSet
>>>>>
>>>>> The nxlucene.rss package is a small package with no dependencies on
>>>>> nxlucene.
>>>> Yes. This is a NXLucene package dependency.
>>>>
>>>>> I would very appreciate it if this package could be brought into the
>>>>> nuxeo.lucene package so that there are no parts that are not ZPL -
>>>>> since
>>>>> I work with ZPL pretty much exclusively.
>>>> right. I don't see any blocker for that.
>>>>
>>>>> Alternatively, if this sub package is needed by both packages, perhaps
>>>>> it could be split out under the ZPL so it could remain in common
>>>>> without
>>>>> affecting the license of nuxeo.lucene.
>>>> yes. this is the thing to do but you'll do the package split your self
>>>> :) Do you have an svn account ?
>>>>
>>>>> Of course, it would be great if you could consider ZPL for nxlucene
>>>>> also
>>>>> - to make these two packages consistently licensed - which would also
>>>>> make me a happy camper. :-)
>>>> Nope. NXLucene will remain under the LGPL.
>>>>
>>>> Cheers,
>>>>
>>>>     J.
>>>>
>>
>>
> 


-- 
Julien Anguenot | Nuxeo R&D (Paris, France)
Open Source ECM - http://www.nuxeo.com
Nuxeo 5 : http://www.nuxeo.org
Mobile: +33 (0) 6 72 57 57 66

_______________________________________________
cps-devel mailing list
http://lists.nuxeo.com/mailman/listinfo/cps-devel
signature.asc (application/pgp-signature, 252 B)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFFYJLJGhoG8MxZ/pIRAnrNAJ4uKmy+r45tzl+NI6/Yev/r9h1IrACfSBcI
W2FFCumCRqou7z6F+YdPL2Q=
=qewy
-----END PGP SIGNATURE-----
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.