Notes from the 2017 Debian Med Sprint in Bucharest

Progress on meeting goals

=== Done ===

• Updated cwltool & schema-salad: done. Added cwl-runner via update-alternatives, also for toil

• Packaged Toil resource distributor for CWL workflows (python-bd2k and essential modules of cgcloud with it)

• seqan 2.3.1 is packaged & uploaded

Ongoing

• Preparing package for conda
• Complete dependencies for Galaxy
  • reiterating on borked python-bx packaging - now in New Queue
  • Remaining missing python dependencies (to be validated by all):
    • testfixtures (needed only for testing)
    • pulsar-galaxy (is it really needed? it is a runner, so could be disabled/removed from package)
  • Galaxy fails with python-routes/python-webhelpers from Debian, more recent than galaxy requirements. Would need patching (but what/how?)

Edit conflict - other version:

• Packaged Toil resource distributor for CWL workflows (python-bd2k and essential modules of cgcloud with it)
• Preparing package for conda

• Updating cwltool & schema-salad: done. Sascha is adding cwl-runner via update-alternatives

  • uploaded

• packaging bcbio: MRC working on it and pybedtools dependency. Bug in underlying bedtools software found: https://github.com/arq5x/bedtools2/issues/468

Edit conflict - your version:

• Galaxy status: several dependencies are not yet python3 compatible and will impact Galaxy code (https://github.com/galaxyproject/galaxy/issues/1715). Upstream is working on this, but no roadmap yet. No python3 support is blocking. Auto install of dependencies by Galaxy in a virtualenv, shared with installation of toolshed means code modification in Debian to use system libs. This *may* be managed with --system-site-packages option in virtualenv creation.

• run.sh options --skip-venv --skip-eggs --skip-wheels seem to disable virtualenv and packages installation. Main issue seems to be samtools-related with imports like import net.sf.samtools.SAMFileReader

• packaging bcbio: MRC working on it and pybedtools dependency

End of edit conflict

• snpeff: looked at dependencies needed to package it

  • apfloat => source at http://repo1.maven.org/maven2/org/apfloat/apfloat/1.6.3/

  • charts4j => from https://code.google.com/archive/p/charts4j/, code at https://github.com/julienchastang/charts4j

  • akka-actor => source at http://dl.bintray.com/typesafe/maven-releases/com/typesafe/akka/akka-actor/2.0.1/

  • trove4j => in debian (libtrove-java) freemarker in Debian

  • distlib => unknown source , only bin jar on maven, but no reference in code...

  • samtools => v1.107, too old, code not compatible with htsjdk...

  • log4j in Debian

• Packaging lobSTR workflow: Bogdan & Michael got the workflow to run again https://github.com/common-workflow-language/workflows/pull/126#issuecomment-272632458

Project ideas

List of demands for development discussed/identified during the meeting

• bash autocomplete from CWL tool descriptions
• manual page generation from CWL tool descriptions

• using acd2cwl to generate CWL descriptions for the EMBOSS tools

• [simple] add "Files-Excluded" to dh_make template of debian/copyright
• patch uscan to allow for a recursive grep to determine what files to exclude via Files-Excluded

• Demonstrate usage of http://snapshot.debian.org/ in a Dockerfile for reproducible computing

• Develop strategy for automated package updates in a 'pantsonfire' distribution