LSC Data Analysis Software Working Groups

Agenda: Wed, 09 April 2008

John Z. to review the NDS2 api that he circulated to the working group earlier
today.

Scott to go over the presentation that he prepared during the face-to-face.

Other than that, the usual round-robin of the projects.

Minutes

Attendance:

Xavier Amador
Stuart Anderson (Chair)
Warren Anderson (Minutes)
Kipp Cannon 
Jolien Creighton
Britta Daudert
Filipo Grimaldi
Scott Koranda 
Ed Maros
Adam Mercer 
Jamie Rollins
John Whelan 
Igor Yakushkin
John Zweizig

Minutes:

Scott - short presentation that there was no time for at DASWG f2f. Slide
posted on DASWG page at

LDRstatusLVCmeeting-03-19-08.pdf

Status and plans for LDR. Graded LDR in various categories. Listed most
significant S5 problems. Listed S5 success stories. S6 can maybe have 30
minute latencies but will never be realtime system. Listed recent
developments. List of development plans for S6. List of why Globus WSRF will
be new framework. See url above for specifics. 

Stuart wonders if WSRF would help solve protocol layer problems for NDS2. 
Scott thinks not - not good for high throughput. For LDR, WSFR will do
planning work and then gridftp will do the heavy lifting. Cost of serializing
xml on both ends would be too great for NDS2 in Scott's opinion. Scott will
use C WS container when needed, but much will be done with Java container
which is much slower. Stuart wonders if division of planning and metadata for
NDS2 would work as well. Scott thinks it might be possible, but doesn't know
for sure.

John Z - NDS2 client API. Set of C functions to use to get data. Based on what
Alex had previously - this way backwards compatible to extent possible, but
original was crude. John has improved some aspects like defining channel
orders, etc. i.e. add intelligence at server end. Most important extension of
the NDS API is channel request list concept. Can now manipulate channel list
by adding or removing channels on the fly. Channels names somewhat dynamic, so
ability to list channels now can take time argument to get channel list for
relevant periods. NDS2 will have ability to mix data from variety of sources -
RDS, online, minute trend, second trend. NDS protocol restricted to one data
source at the time. Backwards compatible channel list will give online
channels at current time only. Documentation at 

group__daq2__access.html

Some of the documented items are more relevant on server end - John wants to
split out just client aspects. Gives item-by-item description of classes in
NDS2 access functions. General set of functions inherited from NDS. Some minor
changes from pure backward compatibility are necessary. 

Stuart wonders if people in design group have given input. 
John says no responses so far. He discussed design with Alex up front. 
Stuart asked who list is: Alex, Josh, Greg, etc. Encourages John to start
soliciting feedback. 
Warren asked about security. 
John says based on discussions will use kerberos. 
Warren suggests GSSAPI as basis.

Projects:

DMT - John says not alot going on - implementing FFT IFR. Speed up over
convultion order of magnitude slower than expected in some cases. Expects it
to be useful for h(t). Stuart suggests contacting Xavi and Greg.

GLUE - Kipp nothing to report. Stuart wonders if ihope lives in glue? Kipp
thinks in LALApps, but pipeline.py which sits under is in glue. Steve
Fairhursts improvements in LALApps.

LAL/LALAPPS - discussion of using libxml2 for xml IO. Do we want our software
dependencies to include this external library? Stuart wants to know who is
pushing? Jolien says most pulsar group, some inspiral. Not LIGOLW xml, so
metaio won't work. Reinhard is doing most work on investigating it. 

LDAS - Ed upgrading framecpp to current specs. Approved update to frame spec,
but not implemented yet, so we can still push some changes. Considering per
frame checksums, more compression modes. Jamie wonders where spec is posted.
Stuart will send email with pointer soliciting comments.

LDG - packaging group led by Scott and Greg. 

LDN - nothing

LDR - nothing

LIGOTools - nothing

MatApps - yet another version of matlab available.

Metaio - nothing

Onasys - job database problem is resolved by removing column whose name is a
reserved word. Scott has done it and updated mysql at both sites. Has to
modify grant tables so onasys user can read and write and Stuart will assign
rebooters for system. Kipp has updated software, no impact on current users.

Online Running - ready to go again.

OSG LIGO - making changes to ihope to run parts of job on OSG. Cannot run all
of dag yet, either modified or original. Britta will solicit help if she can't
figure it out herself. Stuart reports that there is a lot of interest on OSG
side to get LIGO users on OSG.

Other business? None.

$Id: 080409.html,v 1.1 2008/04/09 20:29:34 warren Exp $