2010-05-04 Tuesday Meeting
Attending: Brian, Scott, Colin, Andrew, Chris, Donald, Will, Jean-Marie, Jason, Josh, Melissa, Carlos
Agenda
Remember: Agenda must be complete (with estimated times) on the day before the meeting. Any additions after that must go at the bottom (AOCB)
Accepting minutes from last meeting
Matters Arising (<10 mins)
- Put in action points from last Thursday's meeting.
- Brian: Find out if Michael is available for next Tuesday ASAP
- Brian: Rough testing script for next week
- Brian reply to QA aperio tickets (#2412)
- Chris reply to QA (#2403)
QA and User Testing: Brian's presentation (20 mins)
FLIM: Donald's presentation (10 mins)
Model Changes (10 mins)
- Eveyrone: Make a decision next Tuesday about interative model changes
AOCB (<5 mins)
Notes
Minutes accepted
Matters arising
- QAs replied to
- QA 2403 outstanding. retrieving Leica file
Paris
- meeting done
- Johannes and Pavel attending
- Andrew working on final program
- to-be-confirmed for one title
- @Make page live
Visit from Christoph and Ingvar next week
- They should discuss with Angie about accomodations
- Discussing Java volume viewer integration (Insight & Web)
- Status update:
Possibly attending OSSWatch meeting
- Scott will provide details quickly
Testing presentation / Brian
- State of where we are
- Very little repeatability of our tests
- two pairs of eyes: at least at the client level
- matching external developers as testers with internals devs
-
technical lead for maintaining the design of what we've decided
- best practices maintained by that person
-
continuous integration for knowing what breaks
- but don't have that at the testing level
- "TEST" tag for tickets?
- using videos as the test suite
-
Q/A
- Scott: disagree with the separation
- Jean-Marie: User story is roughly this definition
-
Brian: less repeatable test but tests attached to the user story?
- Scott: Yeah.
- Jean-Marie: need concrete tests (reality is we have no testers)
- Next point: formalize the set of things we test
- Will: that's why the movie is a good starting point
- Chris: watch a movie and reproduce, or spend 30 minutes to write it down
-
Will: what actual problem are we trying to fix?
- bugs (4.1 release wasn't too bad)
- duplicating testing?
- Brian: user perception - we don't do a lot testing
-
Jason: have own memorized
- Why don't we just write down the scripts?
- Jean-Marie: I think that's what needs to be done.
- Jason: wasn't able to keep up with the development
- Didn't know what was accomplished
FLIM presentation / Donald
-
very complicated workflow currently
- generates an Excel spreadsheet for each ROI which end up being compined
-
Jason: no guarantee that each cell is the same
- Differences in lifetime measurements may be real and shouldn't be averaged out
- Donald: want to add statistical analysis to test fit
-
New workflow:
- uses omero ROIs and keyword
- runs a script against the dataset to get 5 spreadsheet results (k1, k2, a1, a2, chi)
-
proposed next additions:
- combined results
- generate histrograms
- generate heatmaps
-
Basic workflow done, but a few problems
-
Polygons not a mask (Andrew would like to scream) [Blocker]
-
Josh: perhaps we could use vertex arrays?
- Donald: can still store polygons (have to be fixed) * Doesn't work properly in measurement tool (bez. issues)
-
Josh: perhaps we could use vertex arrays?
-
Extended workflow, allowing user ability to add steps
- Andrew: just use coloring/labelling of cells?
- Donald/Chris: want something more formal (set keywords)
- Jean-Marie: like ticket fields in agilo
-
Big issue: NDIM
- need time and channel information
- examine it in Paris
-
Polygons not a mask (Andrew would like to scream) [Blocker]
-
Future
-
install.py / upload.py for defining the namespaces, etc.
- i.e. allowing scripts to put scaffolding in place for UI
- cluster deploy
- reusing for FRAP
- better quantification (stats, LDA, fisher score, etc)
-
install.py / upload.py for defining the namespaces, etc.
-
Q/A
-
Jason
- name "roi script", to stay away from the cluttered names
- slide 3 (black v. red): David's response? * He liked it. Will he use it? Can he use it? * Can't use it without polygons. * We've got SPCImage but in OMERO without polygons * To be better: needs to have the stuff in red
- If it's more or less the same amount of work,... * what's reasonable for us to expect red stuff for David to use? * Time scale? 2 weeks to get something in front of him * Donald: have done the hard work already * polygons & the math are a wee bit tricky (couple days of work) * Showing it in Paris? Definitely. * Chris: how hard to get it merged? A day of work. * Will: release as 4.2? Hopefully all of it. * Jason: syncing earlier rather than later is usually a good idea * @Donald: commit by the end of the week. Hopes David can use by then * Chris: releasing this as a preview with a flag
- @Chris: David needs a box to run this (by week's end?)
- Pieta also wants to do her divisional talk on the scripts
-
FLIM issue (mathematical)
- Discussion with Kevin/Curtis re: Paul Barber at Kings
- Rewriting math around flim analysis
-
Jason
-
very complicated workflow currently
Model changes
- Currently: 2 big releases of the model per year
- Example: SPW timepoint(int->timestamp)
- Chris: stripped 2 properties out of mask (width/heigth)
- Itermediate release or do we make changes for 4.2?
-
Andrew: changes would require a major release
- but it's only been out there 1-2 weeks
- Chris: another version at Paris?
- Jason: @but do have to write community now.
Actions
- Andrew publishes the page
- Jason writes to the speakers for info
- Jason writes to Angie, and CC's EBI
- Scott sends details on OSS meeting
- Brian: Start the ui testing scripts for most tasks
- All: send new name suggestions for "workflow" Donald
- Donald commits FLIM code for the end of the week
- Chris: David needs a box to run this (by week's end?)