Attendees:
Fabrice, Daniel, Serge (later)
Status:
Daniel: xrootd port probably finished (working, first pass), but
working with fabrice. Also working on new worker result handling.
Fabrice: Working on non-scons configuration tool (DM-622)--in
review. Non-internet installation (DM-602), blocked because Anaconda
python install script needs internet. Also working on DM-930--
improving documentation as part of DM-622. May take over DM-649,
minimal documentation framework. log4cxx, need to repackage (without
changing the version), need to ask LSST what the policy is for
updating a package's packaging without updating its upstream software
version (1.3.3 --> 1.3.3-1?).
Serge: was sick:not much progress. RefMatch probably won't make it to
June release.
eups packaging and distribution:
One problem: no way to share permissions so that multiple people can
use the eups distribution tool to manage distribution
packages. Suggest a shared account (e.g., qservsw) that the devs can
share. Perhaps use a sudo-like system so that each dev can run eups
commands as that user.
Fabrice thinks that if we make new xrootd packages, it should not
affect installing the 2014_05 release. Current xrootd package
(qsPatch5) is versioned qs5, but eups dislikes this format and
displays nothing for its version when listing packages. Suggest naming
the old one: 3.2.8-qsPatch5 . The new xrootd package will be named:
4.0.0rc3-qsClient2 .
Anaconda: need LSST-approved/supported way of installing w/o
internet. See:
https://dev.lsstcorp.org/cgit/LSST/external/anaconda.git/tree/ups/eupspkg.cfg.sh
Suggest requiring all eups dependencies installable by only
touching the eups distributions server (i.e., not continuum.io
<http://continuum.io> server)
June release: Jacek, please just cut the release from what's already
merged now. Don't worry, July release is likely to contain major
features: refmatch, xrootd port, result plumbing, config tool.
mysql result plumbing: Creating custom local_infile handler in
mysqlclient seems to be the fastest way to get in-process data through
a mysqlclient into a mysql table. Other options: "prepared" insert
statements with binding parameters; writing directly into
myisam. Prepared statements are probably slower--mysql doc "8.3.2.1
Speed of INSERT Statements" suggests that LOAD DATA INFILE is the
fastest way to bulk-load data.
########################################################################
Use REPLY-ALL to reply to list
To unsubscribe from the QSERV-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=QSERV-L&A=1
|