Attendees - Gaines, Berman, Canal, Kennedy, Li, Greenlee, Chlebana, Kallenbach

We will attempt to reserve a room from 1 - 3 each Tuesday, and make use of however much of it we need for that week.

CDF stated that the PASREC charge had been presented at their upgrade meeting. Slides are on www. For now they anticipate that they will continue to use PAW, while getting the new experimenters started with ROOT. They have posted the survey slides on the WWW, and are awaiting comments.

As for platforms, there is a small NT following among the on-line group. Support for X terminals will be necessary. They will have a draft survey ready next week. For now their computing platforms are Irix, linux, NT.

D0 - Online require NT, linux, DEC/Unix. Offline require these plus IRIX. For the long term Sun, HP, AIX may also needed. For data size, they will need to handle 200 histograms x 3000 channels - approximately 15TB. Parallel processing (sample data from multiple processes to 1 histogram) required.

The D0 representatives stated that they would like to make final choice, not an interim one. They are willing to wait until 12/98 for a permanent solution. At the present time they are using PAW and histoscope. Once a permanent solution is recommended, the various users will switch when appropriate. As for the functional breakdown in the requirements document, D0 considers simulation + reconstruction to have the same requirements.

PAW - let die out

ROOT - The ROOT user interface is different than PAW one. The ROOT ui is deficient, in that commands are not intuitive. Training classes will probably be needed to familiarize experimenters with the command set. Tutorials and other documentation are on www. At present, the data model is limited. Support for cint is basically non-existent. A PAW-like support, including fulltime local support, will be needed. The present status of the product makes it a good candidate. RK, PM

LHC++ - 1^st production release imminent. However, only the HP and Sun versions will be available. Ports to our preferred platforms are unclear. The many product dependencies involved with LHC++ make this problem worse. For example, objectivity is needed, and no firm commitment to Linux is evident. The product structure, being comprised of many commercial components, will make limitations on component products propogate to the whole package (for example, specific compilers are needed for some components, and KCC is not supported). Even though a full release, including Win32 and Linux versions, is expected in Q2 99, we don't think LHC++ should be evaluated at this time.

Nirvana/Histoscope - Certain deficiencies are being fixed. For rxample, the program now runs on Win32. There is a narrow user base for now. "Important functionality", such as a scripting language, is missing. We would like to evaluate this product, working with the Nirvana development team to fix the deficiencies.

Can shareware (e. g. Python) fill holes?

MATLAB - The primary commercial candidate. Concerns from PASFRG and PASSUMA will have to be investigated. These include cost of licenses, upgrades, support (access to source or turnaround on bug reports), and handling of our large data sets. It was also pointed out that algorithms may be different from ones in our current software. As the most widely used commercial scientific analysis software, MATLAB should be evaluated.

Octave - The freeware version of MATLAB. Because of its MATLAB functionality, this product should also be evaluated. Support issues peculiar to commercial software should not be present.

7/14 - Surveys, reminders, first initial reports, review plan

7/21 - ROOT, Nirvana (w/ Paul)

7/28 - MATLAB, Octave

Links to candidate products will be added to PASREC page.