[DAS2] DAS/2 weekly meeting notes

[Chervitz, Steve]

Notes from the weekly DAS/2 teleconference, 21 Oct 2005.

$Id: das2-teleconf-2005-10-20.txt,v 1.1 2005/10/21 23:55:05 sac Exp $

* Gregg and Lincoln hashed out some details regarding specific aims
  for the DAS/2 continuation grant, refining these aims:
   - Client enhancements
        + includes client-side caching
   - Spec enhancements
        + caching (IF_MODIFIED_SINCE support)
        + dealing with lengthy server operations
        + extensions to stylesheet capabilities
        + embedding arbitrary snippets of XML (e.g., RDF)
   - IGB-Apollo interactions
        + Gregg has been doing some background research on HTTP and
          UDP protocols

* Greg plans to have a draft ready to circulate for review by the
  participating organizations by Tues next week (10/25).

* DAS/2 meeting at CSHL Genome Informatics meeting:
   - Planned for 10/28 evening (7:30pm EDT) in Lincoln's office
   - Teleconf will be available

* Proposal to move the weekly DAS/2 meeting to 9:00am PDT/PST Thursday
  so that UK folks can participate
   - Folks will check schedules to see if this works
   - Chime in if it doesn't work for you

Status Reports
--------------

Gregg:

* Completed treeview of DAS/2 sources in IGB. See his recent post with
  GUI snapshot. This will be included by default in the next
  IGB release next week, to coincide with the CSHL GI meeting.

* Invited to talk at CSHL GI meeting on genome viz. Good opportunity
  to present IGB. May accept the invitation if it's later during the
  meeting. 

* Is making an effort to respond to questions on the DAS/2 list.

* IGB release planned for next week, coinciding with CSHL GI meeting.
   - Lots of new additions (DAS/2 support, new GUI preferences
     setting). Might be a 4.0 release.
   - Access via Java Web Start
   - Users will be able to launch previous version of IGB, if they
     have problems with the new version.

Discussion:
   - Ann Loraine puts in a strong plug for JWS-based application
     launching. It's just so convenient, even for someone who is
     comfortable building and running java code with command-line tools
     (but doesn't always have time to do so these days).

   - Gregg would like to set up an auto-build system for IGB code on
     SourceForge (or open-bio.org, if SF doesn't support this). This
     would allow users to download a pre-built jar. A JWS system could
     then be built on top of this if we want.

Ed E:

* Lots of IGB commits last week.
   - Can now set persistent preferences via GUI rather than pref file
     (bg colors, feature colors, number of levels of a given tier,
     show/hide prefs, etc.). Much nicer user experience.
   - No documentation yet.
   - Gregg adds: Deb Coulton (sp?) is now revising the IGB user
     guide. Targeting end of Nov for release.

Steve C:

* Working on improving the DAS/2 spec, resolving bugzilla issues,
  consistent wording, etc.
   - See http://biodas.org/documents/das2/das2_get.html
   - Be sure to force your browser to reload the page daily to make
     sure you have the latest version.
   - Will post notable changes to the DAS/2 list.
   - Will post any issues requiring broader discussion to list before
     modifying spec.
   - Gregg adds: See his response to Andreas on the list regarding how
     to deal with server agreement on what sequence coordinate system
     is being used. Do folks agree with Gregg's proposal?

Allen D:

* Bug fixes on biopackages.net DAS/2 server
   - A DAS2XML parser would help catch bugs in advance. Will ask
     Andrew about setting this up.

* Discussed performance issues.
   - Feature request is as fast as it's going to get now (in terms of
     SQL optimization).
   - Has a SQUID proxy set up to help with relatively static data, but
     you must configure your client to use it.
     (also ss Allen's post on 10/21 regarding aache mod_cache module).
   - Gregg adds: Would like to get support in spec for 'if modified
     since' tags.
   - Gregg: What about using a different schema underneath, such as
     GFF-db as used by gbrowse which is quite fast?
   - Allen: It's not fast for large segments. The DAS/2 server uses a
     chado schema with feature indexing optimizations by Allen. It's
     performance improves for big queries, but degrades for small
     queries. So the DAS/2 server's response is more flat over a range
     of query sizes relative to gbrowse.
     Possibly could improve the DAS/2 server by upgrading from
     Postgres 7.4 to 8.0, which has improvements on left joins.
   - Lincoln: How could gbrowse performance be improved?
   - Allen: Via partial indexing. Should improve for queries covering
     200+ MB.