Attendees: Daniel, Bill, AndyS, Serge, AndyH, Jacek

The whole meeting revolved around discussing the
Qserv 2014_05 sprint: http://ls.st/bwa


AndyH
-----
DM-220
  - Daniel keep finding little things
  - Andy pushed new changes/fixes over the weekend
    - not reviewed yet
  --> keep this issue open


AndyS
----
DM-613
  - blocked by several issues:
     - ORDER BY, DISTINCT, different column names
     --> Andy will open ticket about DISTINCT problem
     --> mark the blockers in DM-613

DM-228
- close it

DM-531
  - consider converting all arguments to string
    - could be problematic for nulls
  --> Andy will push code to branch, Jacek will review

DM-546
  - Daniel will review soon


Bill
----
  - great progress with logging, most features requested
    by K-T, Robert, and Jim implemented
  - A partial list of what is finished:
    - Refactored entire solution to achieve more simple
      design (e.g. removed a couple of ugly switch statements
      and the need to pass around function pointers).
    - Can retrieve a logger object in C++ layer for
      high-performance applications.
    - Formatting in python layer is now deferred until
      after threshold is met.
    - Programmatic control and querying of logging threshold
      in both C++ and Python layers.
    - Implemented context objects for both C++ and Python
      layers, which automate pushing/popping of names onto/off
      of the global context using constructors/destructors.
      In Python, the context class can be used with the `with'
      statement or as a standalone
    - Re-worked the Python logging module ProtoLogHandler.
      Fixed bugs. Now more robust.
     - Implemented boost::format interface as Robert requested
      (a big change).
  - Also investigated the introduction of new logging levels.
    This is more trouble than it's worth. It would significantly
    complicate the implementation, we can use multiple loggers
    to achieve anything that can be done with additional logging
     levels. See example on trac to illustrate this.
  - To do
    - update the online documentation.
    - package protolog.


Daniel
------
  - most time spent on finishing the XLDB 2013 report
  - DM-633
  - Daniel away tomorrow


Fabrice
-------
  - DM-612 is ready for review, AndyS reviewed,
    Jacek will review very soon


Jacek
-----
  - @xldb-healthcare in NY Mon-Wed
  - several cleanup type issues implemented/done
  - DM-508 and 313 - Daniel will review soon


Serge
-----
  - not much time for qserv this past week
  - have plan for refmatch
    - same logic as for spatial join
    - there is only one small special case: need to
      remove duplicates if we query just match tables
    - also need to tweak join code, it currently does
      not look at join conditions (needed elsewhere,
      not just for refmatch)
  - refmatch related code = ~200 lines (guess)

  - DM-626 - Jacek will try to look
    - code is related to bug with DISTINCT


Non-sprint topics
=================

unit tests not triggered during build
  - we used to have it prior to W14, disabled
  - want to have it back
  - need new code (we have new build system)
     --> open ticket, assign to AndyS

permanent location of how-to-run-qserv?
- want it to be always in sync with master
  --> keep in the code repo, in some common
      format, maybe markdown, and expose it
      on qserv page
    - Daniel will look into it

Migrating Qserv trac pages to Confluence
  - try doing itfor Qserv pages (Jacek will try)

Dev meeting a.la. how we run the Dev meeting
last week
  - 4 hours, every week
  - Wednesdays 12-4 pacific, replaces Qserv Thu meeting

Jacek

########################################################################
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