Agenda

• update on XSLT fun (Group A)

• QA

  • status
  • todos

• Permissions

  • Todos
  • Priorities

• Analysis Preview

  • Rois : Ellipses!, Masks?
  • Measurements : service, parsing
  • Formats : directory import, dropbox

• Data Model

  • Filter, detector proposal

• 4.1 schedule

  • Graffle : usable? offmark?
  • iteration 1 : remote dropbox, rois, permissions, validator (on track? date?)
  • iteration 2 : codegen, xslt, qa, ... what else? starts?
  • Missing from plan
    • DB Upgrade : new formats? freeze?

• forums and emails

Notes

• xslt (Andrew)

  • transforms in place that work for given data
  • 2003, 2007, 2008-02 --> 2009-09
  • don't have tests written
  • broken: roi in source file is rubbish, get gibberish out
    • still written under limitations
    • loop options for z&t --> stuck in continuous loop
    • need tests for that
  • another week for all of them?
    • all transforms done.
    • write tests, fix roi
    • need help? probably. (input/output files validation)
  • reusable chunks: need to refactor.

• QA (Ola)

  • metadata-validation collector is done
    • checks all required/optional(linked) fields
    • doesn't do this via OME-XML but via models.
    • next phase is to check the data
      • not checking the fields currently
    • currently checks timestamps and structure (proper links)
    • Chris: don't worry about displaying the output?
      • Ola: perhaps a simple display to know we're collecting properly
      • parsing XML file with 2000 record is ok. how fast with 10K? more
  • need testing with Chris, but generally works.
  • need more time for displaying the collected data
  • currently limited by a memory leak
    • can only import about 1000 files
    • @Brian will add a ticket
  • cF. previous validator was just for XML/TIFF
    • two different things
    • review when OME-XML generation is always valid
  • Curtis: providing these tools to do all these checks
    • warnings for LSID, etc.
    • Adding xml generation to metadata-validator calls!
  • NB: should be straight-forward to take an unordered XML
    • and produce an ordered XML
    • Will: or removing the sequences?
    • Curtis: Xslt would be fantastic (short to mid-term)
  • Moving comments/bugs to new system?
  • data collection: only we will be able to see data.
  • Will: moving all the existing feedback?
    • Not migrating everything. Starting fresh.
    • Everything turned into tickets? Yes.
    • Linking between trac and qa. (have a separate trac for testing)

• permissions (Brian)

  • private space, and a single or multiple public space?
  • fine for vast majority of people
  • private space gets rid of PI concept. perhaps via 600
  • Donald: no private space?
  • Brian: can you add more than one boss?
  • Chris: how does this work with the setting a group to public from private?
  • quickest option:
    • uesr only in one group
    • turn group private or public
    • server ensures graph-consistency, both group_id and permissions
    • PI or admin can move group to public
    • UI: need to know group prespective (don't show data)
    • ADMIN UI: need to allow upgrade to group visible. can't downgrade
      • offering upgrade button
      • sending email to all the users??
  • next stage: either private space or multiple groups.
  • really a testing issue: moving the component/client tests.
    • get them green
    • then modify them as expected

• analysis preview

  • import (Chris)
    • getUsedFiles is populating, only on images themselves.
    • big hash table is now associated. (great!)
    • Questions:
      • two hashtables: global metadata & series metadata * combined together, since many don't have series * a bit of data duplication. shouldn't do any harm.
      • every image will now have an annotation (yellow flag) * do companion files count as annotations? * already ruins the use. * ignoring file annotations? No. That's not good.
    • adding relevant file formats to db
  • rois
    • companion files have roi data (geometetries)
    • overlay files: masks
    • ellipse rois now displaying in the client
    • measurement tool will be read only.
    • extremely slow to get from the server.
    • mias
      • 2 sets of data. tab separated rois (assumed circles)
      • tif file overlays. rois drawn in red/green/magenta/...
      • no correlation between the two
      • adjusting pixel infrastructure with bit depth * take each color as a mask * show in rendering engine as channel * ..or.. insight as an overlay
      • files have a color map; last five elements in the color map
      • parsing it out is simple.
      • overlapping? yes.
      • talk @ 1400-1500 (MAX!) BST for rois
  • dropbox for hcs
    • subclassing file monitor, waiting for no activity or a specific file
    • then messages the monitor system
  • chris: start work on directory-based import next week
  • Curtis: metadata as attachments. Why not as SAs?
    • Chris: we are using SAs. Just a file-based SA. (i.e. performance)
    • XML file later, if we want.
    • Jean-Marie: editor! :)
    • Curtis: easy to have editor ready k-v pairs. ini headers? Yup.
  • Curtis: roi coordinates
    • massaging based on other formats?
    • same case in 2D as 3D?
    • define shapes differently. define in schema.
    • need to continue conversation later. - imaris does things in a particular way - need to review what others do. - making the pixels selected as closely as possible. - two different definitions for each geometry? - massaging/transforms? - well-documented in the schema! (current and future)

• data model

  • filter proposal
    • few people read it. few suggestions from curtis (remove attributes)
    • shorter proposal for filter
  • detector http://www.ome-xml.org/ticket/65
    • is more of the start of a discussion
    • did anyone look at it?
    • Andrew: why do need the light path object?
      • contents could be in channel itself.
    • filterset? very specific thing.
      • is it the term for a type of filter? don't think so.
    • curtis: still a valuable concept. need a place to store that
    • jean-marie: keep filterset and add phase (collection of filter)
      • each linked to channel?
    • everything is duplicated
    • Andrew: putting it on instrument and reusing?
      • everything else is
      • Will: but more like detector settings.
      • filterRef is becoming a filterSettings object
  • planning other meeting
    • proper XML examples
    • curtis attend or review after the fact? will try to be there.
    • 1400-1500 (MAX!) on Tuesday BST

• 4.1 plan updates for week two

  • opinions
    • graffle nice. update it? yes
  • xslt tests, ordering
  • metadata-validator xml-generation tests
  • qa: finishing xml validator, update registry, changing comments over? reviewing UI next week (wed)
  • perms: josh/brian moving tests from components/client
  • evaluating file annotation filtering in counts.
  • file formats in db
  • rois: speed, overlays, definition review
  • dropbox for mias (in progress)
  • directory import (next week)
  • db upgrades
  • new measurement ("Tables") service for beginning of week
  • confocal metadata completion