PASREC - On Thursday people from IDL are coming to give a demo - 9.0 am in 11NW. Commercial analysis package.
Current stage – less interested in commercial products than originally thought. Partly because of licensing costs- don't seem to understand our model of working or hundreds of people. Price themselves out of the market – may an order of magnitude too high, but they are thinking in terms of 20 or 30% discount. Right now what probably rules them out is their dealings with large datasets – set up as arrays of data which you bring into memory.
Don't understand HEP functionality like event loop. ? not clear whether they have back-end database functionality. Fancy graphics on arrays of numbers read into. Maybe a little bit early for Run II.
The 2 HEP standard products looking seriously at are Root and Nirvana. Agreed on a statement characterizing the major features of those products. With a certain amount of work a year from now both would provide you with similar y. Root exists, don't control the source. Nirvana we have source, maybe based on a more forward looking strategy of components,etc. but functionality a bit behind. Could be risky also people want to use products now and prepare packages for Run II. Hope to have some tentative recommendations by Sep. 1.
Not clear how well people on the committee represent their constituency
– hard to get the word out. Uninformed masses would just vote for PAW.
Analysis Hardware Working Group had its first meeting. 2nd meeting this Friday.
Getting lots of Linux machines as prototype farms. Who is thinking about clusters of small boxes for Analysis? Technology discussion of can this fill the needs are ongoing in this forum.
Prototype Farm system procurement
moving ahead this year.
15 dual processors + 3 i/o nodes split between CDF, D0 and the theorists. Instead of CPS – use a batch system. Also meeting with experiments and E871 getting hands dirty.
Enstore – working on error control. Getting a design which allows retries. Tomorrow break down into tasks to go off and implement.
Requisitioned PowderHorn upgrade and Redwood. Only 2 tape drives. Will have to be shared.
RIP – hardware part of RIP – have networking hardware coming
In and have started working on gigabit Ethernet. Some of the Fiber channel also in. Specification for the data logger, meetings with experiments. For D0 straightforward and will use enstore. For CDF depending on their architecture for network - will see what they do. Stephan has drawn out a disk which can be shared among data logging tasks and analysis tasks and then something writes from this disk to tape drives. So something about allocating
The tape drives issue is intertwined with data access model?
ZOOM - more like C++ working group. Experts from Kuck coming tomorrow to entertain questions about why recently taken a right turn off the bridge in terms of linking and template strategy. Claim that to satisfy new ANSII standard forced into "an awkward mechanism" at best. Issue for SoftRelTools people to be concerned about. Also other questions people have. Still slow compiling – anything they can do about this. Effort going on with pre prototyping browser in histoscope – for columnwise entuples
Get most important features on the table early.
Tools. Liz sent mail about purify. CDF copy disappeared, not properly maintained as a product. Latest copy is on their web site,
We had a beta – now a version. How do we get this?
Chris Green – doesn't want to maintain.
Visualization- all OI licenses we purchased are deployed. MINOS has one. ?? should we be buying more. Is the year up?
No final decision in CDF – group which expressed interest in doing display is leaning towards. What about the linux port?
Tested it?
Simulation – keep working on mcfast. Ready to allocate manpower for STDHEP – figure out what we do with this.