Next Meeting:
Thursday 5 April 2001 at 10AM PST

Number of LAM processes should be allowed greater than three
Phil claimed that during the MDC the MPI API can start with any number of nodes. But he was not clear about the minimum number of processes that were allowed. Duncan will investigate this and present results at the next meeting.

Pipelines in LDAS
Phil will close out PR 791. Phil also reported that a single frame was read into frame API, passed to the datacondAPI, computed power spectrum, passed it to the wrapper API, executed the trivial DSO, sent output to the event monitor and on to the metadata API. Great work. Phil will find the name of the ramp channel and verify that the correct data is acquired by each API and will report on it next week. Teviet and Duncan will produce a "not-so-trivial" dso which generates some output to be passed to the event monitor for verification purposes.

Event monitor API
Masha has been putting 1/2 her time into the data pipeline, so she had to make changes to the wrapper API to be able to read input from file and data socket at the same time. Event Monitor API currently collects data from wrapperAPI job, when it gets the last element, plits it into 3 new elements depending on multi-dim-data, database, state information. She is working on the database translator.

Stochastic background DSO
Warren has taken the lead on implementing the stochastic search in LALWrapper. Nothing much has been done yet. He will implement the four functions with comments and dummy LAL functions. Warren is confident that he can do this by the stochastic inch-pebble MDC.

Patrick raised possibility of wanting a special datatype for detector information in the wrapperAPI. Kent drew line in the sand; no search specific datatypes will be included in the wrapperAPI. He'll go for this only if the information is generic and needed by all searches. We will revisit this issue when the LAL and frame data types have been frozen.

FCT DSO
Linqing is almost finished writing the skeleton code for this. It follows the power DSO model. Some minor differences. She is happy that they will be able to implement the FCT in the wrapper API.

Duncan mentioned that the problem with ssh and "searchXX@....: permission denied" error seemed to show up again during the LDAS demo at the LSC meeting. He asked if there was any progress moving the home directories on the Beowulf. Kent said that we should bu Stuart about it. That seems like the only way that it will get done.