[Mulgara-dev] dateTime information being corrupted

[Chris Wilper]

It may be a longshot, but there was a date format related issue a while
back, which has been fixed since Kowari 1.0.5:

http://sourceforge.net/mailarchive/forum.php?thread_id=6710093&forum_id=43571

This one had to do with concurrent use of DateFormat objects within
Kowari.  It ended up creating incorrect dates when we saw it occur back then,
but I'm not seeing how your described use would provoke it, even if you were
running on a version without the fix.

- Chris

-----Original Message-----
From: mulgara-dev-bounces at mulgara.org on behalf of David Moll
Sent: Thu 11/2/2006 1:06 PM
To: Paul Gearon
Cc: mulgara-dev at mulgara.org
Subject: RE: [Mulgara-dev] dateTime information being corrupted
 
Unfortunately the problem appears to be completely random.  But it is
happening with more frequency.
 
I'm trying to get to the point where we can mirror our production system
with Mulgara replacing TKS.  I believe that we're almost there, with
just a few last details that need to be worked out.  I went through all
of the old issues in Kowari on Sourceforge and I didn't see anything
related to this, but it could be caused by something that doesn't
mention dateTime at all, naturally.
 
Or, it could have been fixed in MGR-2!
 
The only thing that has me hesitant to set up Mulgara on a production
mirror right now is the memory mapping problem, but I don't think that
it is a real issue for us on production.
 
~Dave


________________________________

From: gearon at gmail.com [mailto:gearon at gmail.com] On Behalf Of Paul
Gearon
Sent: Thursday, November 02, 2006 1:03 PM
To: David Moll
Cc: mulgara-dev at mulgara.org
Subject: Re: [Mulgara-dev] dateTime information being corrupted


Ah, OK.



On 11/2/06, David Moll <DMoll at viewpointusa.com> wrote: 

	Actually we're just doing an insertion of raw triples.  The
insert query looks like:

<snip/> 

	This is also the query that we run to correct the incorrect
dateTime elements, and checking the triples immediately after they are
inserted shows the correct information.  So we are reasonably certain
the the problem is not in the initial insert query that puts the triples
into the data store, but is occurring at some later point.


Doesn't sound like an easy one to track down.


Do you have any way to reproduce this?  I suppose it is completely
random?



	Logging is not currently running on our production system.  I
have not looked at the log4j-conf.xml file much as of yet.  Our
production system is running TKS, so while I can't look at any of that
code, I can investigate the logging options that should be turned on the
next time we restart the system.


TKS?!?  Erk.  I never knew when TKS releases were built, so I don't know
which will be your corresponding Kowari.  You may well be looking at a
bug that was since fixed, but I don't know how to find out. 


I'm not about to suggest that you change your production system, but is
it possible to duplicate it using Mulgara (or a recent Kowari) instead
of TKS?  If this is still a problem then we will need to get to it
quickly.  I have a feeling that it may be resolved though... 

Paul


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mulgara.org/pipermail/mulgara-dev/attachments/20061102/957f51d7/attachment.htm>