Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

  1. Notetaker volunteer
  2. Review last week's actions
  3. Status questions/clarification
  4. Review recently submitted issues (15 minutes)

Discussion Topics:

  1. TBDYourKit findings

Individual Status

If you have status items to report, please enter them below at least 1 hour before the meeting starts.

...

  • Released CloudSync 1.0.0 last week, finishing up docs & announcing to list this week.
  • Experimented with YourKit. Did a call timing test (turned on CPU profiling w/tracing during fedora-ingest-demos.sh) and found an obvious area for improvement:

Notes

Attendees:

  • Chris
  • Eddie
  • Ed
  • Steve
  • BenTBD

Notes:

See notes on IRC. Pasted below:

Wiki Markup
\[15:00\] <cwilper> Starting https://wiki.duraspace.org/display/FCREPO/2011-11-01+-+Fedora+Committer+Meeting
\[15:03\] * barmintor (barmintor@specdl11.cul.columbia.edu) has joined #duraspace
\[15:03\] * bradmc (~bradmc@207-172-69-79.c3-0.smr-ubr3.sbo-smr.ma.static.cable.rcn.com) Quit (Read error: Connection reset by peer)
\[15:04\] * bradmc (~bradmc@207-172-69-79.c3-0.smr-ubr3.sbo-smr.ma.static.cable.rcn.com) has joined #duraspace
\[15:07\] * EdAtTheAlliance (811398fe@gateway/web/freenode/ip.129.19.152.254) has joined #duraspace
\[15:09\] <sbayliss> Steve starts taking notes...
\[15:09\] <sbayliss> Actions from last week:
\[15:09\] <sbayliss> * update release availability
\[15:10\] <sbayliss> * have a go with YourKit
\[15:12\] <sbayliss> Release availability page is showing "days" but people are recording hours. Just updated the page to say "hours"
\[15:12\] <sbayliss> Chris reports CloudSync 1.0 is out
\[15:12\] <sbayliss> Chris will go through some YourKit findings later
\[15:13\] <sbayliss> Reviewing recently submitted issues...
\[15:13\] <sbayliss> FCREPO-1023
\[15:14\] <sbayliss> No W3C SPARQL Query Results XML format
\[15:15\] <sbayliss> Would be good to find out what the specific differences are
\[15:15\] <sbayliss> Commented askiing for more info
\[15:15\] <barmintor> <sparql xmlns="http://www.w3.org/2001/sw/DataAccess/rf1/result">
\[15:16\] <sbayliss> FCREPO-1022 Deprecate Messaging client
\[15:16\] <barmintor> w3c root: <sparql xmlns="http://www.w3.org/2005/sparql-results#">
\[15:16\] <barmintor> this may be a Mulgara issue; I think we just pipe those results back (I'd have to check)
\[15:17\] <sbayliss> and FCREPO-1021 Split Java REST client into two artefacts (related)
\[15:18\] <sbayliss> Both opened, assigned to 3.6
\[15:19\] <sbayliss> Eddie to add Steve to Github repo (ID: penthes)
\[15:20\] <sbayliss> FCREPO-1004 Fedora object generator
\[15:21\] <sbayliss> - opened; Frank is doing some work on this
\[15:21\] <sbayliss> FCREPO-1003 - discussed on a previous call, opened
\[15:21\] * EdAtTheAlliance (811398fe@gateway/web/freenode/ip.129.19.152.254) Quit (Ping timeout: 265 seconds)
\[15:22\] <sbayliss> FCREPO-998 - RI Module fails to start if log level is set to DEBUG
\[15:22\] <sbayliss> priority \-> major
\[15:22\] <sbayliss> \-> 3.6
\[15:23\] <sbayliss> \-> open
\[15:23\] <sbayliss> FCREPO-997 - external datastreams: external service MIME type is passed through (not the one defined in the datastream)
\[15:23\] <sbayliss> Discussing if this is the right way or not...
\[15:24\] <sbayliss> Do we want Fedora to override external service mime-type?
\[15:27\] <sbayliss> Reasonable approach seems to be that if MIMEType is not defined in Fedora, then use external service MIMEType. If it is specified then use it.
\[15:28\] <sbayliss> Ben - it's really a feature request, not a bug
\[15:29\] <sbayliss> Current behaviour is probably not well documented
\[15:30\] <sbayliss> Issue Types: Feature is now Story
\[15:31\] <sbayliss> (non-committers may not be aware of this)
\[15:42\] <sbayliss> Would be good to have a custom form so folks get informed
\[15:42\] <sbayliss> Chris will look into this
\[15:46\] <sbayliss> Chris reporting on YourKit
\[15:47\] <sbayliss> Looked at call timings
\[15:47\] <sbayliss> fairly easy to do
\[15:47\] <sbayliss> tested ingesting demo objects
\[15:48\] <sbayliss> example: Schematron validation 95ms in an ingest of 186ms - ie half the ingest time
\[15:49\] <sbayliss> possibly stylesheets compiled every ingest
\[15:50\] * EdAtTheAlliance (811398fe@gateway/web/freenode/ip.129.19.152.254) has joined #duraspace
\[15:51\] <sbayliss> YourKit pretty impressive, Chris found this relatively quickly
\[15:52\] <sbayliss> An issue could be raised for this
\[15:52\] <sbayliss> Q: is this a story?
\[15:53\] <sbayliss> probably it can be written as one
\[15:57\] <sbayliss> eg Epic could be about improving ingest time, Story could be about improving schematron validation performance
\[15:59\] <sbayliss> Chris will create a story for this
\[16:02\] <sbayliss> One for the future: discuss validation, plugabble, disabling, etc
\[16:03\] <sbayliss> Meeting concluded
\[16:03\] <barmintor> thanks for note-taking, sbayliss\!
\[16:03\] <sbayliss> You're welcome\! Made me pay attention

Action Items

...

Eddie:

  • Add Steve to mediashelf/fedora-client git repo

Chris:

  • Look into customizing JIRA issue submission page to articulate what a story is (users no longer see "Feature Request", which may be confusing)
  • Look into upgrading to very latest Greenhopper, which requires JIRA upgrade per Eddie
  • Create story for Schematron validation based on YourKit findings

Steve:

  • Create issue for on-ingest validation configurability (make more relaxed, like disabling schematrong, or more strict, like enabling integrity checks)

All:

  • Experiment more with YourKit. See if you can find other obvious bottlenecks.