Still here and going to New Orleans
January 18th, 2008 by Borkur Steingrimsson
Just a quick post to tell people I am still alive. Haven’t been blogging for close to two months now so I thought I’d better show a small sign of life.
I spent New Year’s eve up in Iceland, visiting friends and family. As soon as the smoke from all the fireworks cleared I took on a 10 day gig with an old client up there and spent my days (as short and dark as they were) sitting next to Pete Scott and drinking coffee. We had run in to a bit of problem with bad performance of some OWB mappings involving slowly changing dimensions. Miracle Benni had already applied a patch that should take care of this (though the patch has been removed from Metalink, replaced with another patch which was then kicked back to development and is finally back now with a description that doesn’t seem to relate to SCD at all ..). We spent a few hours on investigating why this patch had solved the problem when it was identified first but now the terribly slow behavior was back. In the end it turned out that the patch documentation was a tad confusing, to put it mildly. The cunning linguists who wrote the instructions managed to confuscate the readme.txt well enough that whoever applied the patch couldn’t decipher that the patch had be applied on all client machines as well.
I got an email yesterday from the ODTUG folks informing me that my paper got accepted for the conference in New Orleans this coming June. That will be fun, for sure. I am also hoping to attend Collaborate ’08 and although I didn’t submit anything for this one I can always enjoy the work of others :)
As soon as the details have been ironed out, I will be making an announcement here on some upcoming changes …
March 6th, 2008 at 9:18 am
Mark–
Please let us know what happened with applying the patch on client and server. We are continuing our investigation of these issues.
March 7th, 2008 at 4:23 pm
@Antonio
Putting the patch on all of the development client machines allowed us to generate better code (a more efficient date match than previous horrible string match) and allowed us to drive the SCD2 update down from 10s of hours to minutes. We still a problem with SCD2 lookups not restricting to latest values but is probably a different problem.