Skip to content

Instantly share code, notes, and snippets.

Embed
What would you like to do?

Google Summer of Code 2017: Final Report

by Manvendra Singh

Hey everyone! I'm Manvendra, a Google Summer of Code '17 student developer. This summer, I worked on Common Workflow Language Project which comes under OBF Org. This is a summary report on what I managed to do in the past three months.

My major deliverables were:

  • Adding Python 3 compatibility to CWL Tool, Schema Salad and CWL Test projects.
  • Improving mypy annotations and upgrading mypy dependency for CWL Tool and Schema Salad.
  • Reducing issue backlog and misc enhancements.

It was a great experience and I got to learn a lot about python, mypy, debugging effectively and several libraries (both standard and third-party) and a lot of other things. I was the highest committer in Schema Salad and CWL Tool from the start of my coding period i.e. end of April to mid-August: commit graph, commits to master

List of PR's

Here's a list of my PR's which got merged during the coding period.

  • CWL Tool

Link to PR Description
#523 Fix linting and minor errors in README
#519 README: Add more verbose instructions to install from source
#512 cwlrdf.py: Pass str rather than bytes to write function
#509 load_tool.py: Pass to regex before checking for acceptable versions
#505 Add support for Python 3 conformance testing on Jenkins CI
#497 pack.py: add cwlVersion in pack dict, if only single item exists in $graph
#496 Pass group id (GID) in --user flag when calling docker run
#494 sandboxjs.py: Fix check condition for downloading node:slim pull
#491 load_tool: If no version or incorrect of cwlVersion is found, throw fatal error
#477 setup.py: Remove condition which blocks python 3 install
#476 Use io.open instead of builtin open
#470 tox.ini: Remove redundant unit tests code blocks command
#464 Turn on Python 3 on travis CI
#461 Upgrade mypy to v.520
#460 mypy: Don't silence schema_salad errors
#450 setup.py: Pin ruamel.yaml package version
#442 Mega PR for Python3 compatibility
#423 Updates to mypy
#407 Sorting python imports using isort
#406 requirements.txt: Use same version of rdflib as in setup.py and schema_salad
#393 Prevent python3-pip installation until py3 is supported
#383 Added ISSUE_TEMPLATE.md
#337 sandboxjs.py: Check for valid version of node/nodejs engine
  • Schema Salad

Link to PR Description
#128 Bump version of avro for Python 3
#125 Fix windows installation error due to absence of \tmp folder
#123 Integrate avro fork in Python 3 installation
#122 Use io.open instead of builtin open
#121 Update mypy to 0.520
#118 Pin ruamel.yaml dependency
#116 requirements.txt: Add mypy, typed-ast packages, use in tox and Makefile
#114 Minor refactor: Create utils.py
#109 Mega PR for Python 3 compatibility
  • CWL Test

Link to PR Description
#20 Add Python 3 support
#33 Add classname category to specify Python run time version in XML output file
  • Some fun stats

git log --author="Manvendra Singh" --oneline --shortstat
Repo LOC ++ LOC --
cwltool 6,731 9,469
schema_salad 1,965 3,158
cwltest 34 20

Total LOC: 8370++, 12647--

Work done

Phase 1: 7 weeks

  • Made Schema Salad code Python 3 compatible and fixed mypy annotations. The majority of the changes were made in this Mega PR.
  • Worked on CWL Py3 compatibility: Again majority of the changes came in a Mega PR
  • Fixed mypy annotations and upgraded to v.511
  • Ran mypy in Python 3 mode without any errors. This required adding compatible stub files.
  • Dealt with Apache Avro issues
  • Fixed failing unit tests
  • Fixed failing conformance tests
  • Added Py3 compatibility to CWL Test

Phase 2: 4 weeks

I tracked my Phase 2 progress using Manvendra's Issue Tracker. This included addressing very specific issues and enhancements:

  • Better errors for invalid cwlVersion
  • Fix node:slim check and download condition
  • Update Jenkins script to add Python 3 conformance testing
  • Fix Windows installation error
  • Fix packed workflow bug related to cwlVersion
  • Add classname field in XML output report when running CWL Test
  • README: Improve installation instructions and formatting
  • Deprecate CWL draft 2,3,4 support on CWL Tool

Challenges

Some of the major challenges along the way:

  • CWL User guide is rich but the codebase is relatively less documented. A major challenge was to figure a lot of stuff myself and how a lot of code is interacting. Michael was quite helpful and very responsive to queries. Going through testing code also helped.
  • Strings needed to be handled correctly: differentiating between binary strings and ASCII/Unicode strings. Discussed this issue in two of the weekly posts: 1, 2
  • Dependency upgrade: Getting Apache Avro working under Python 3 runtime turned out to be a big challenge: issues/524
  • Fixing failing conformance was also a challenge, but an interesting one. Even though conformance tests cover less than half codebase of what unit tests do, they helped uncover a lot of subtle and integration testing bugs while porting.

Future Work

I've a few Work In Progress PR's currently which I plan to complete.

  • Add contributing guide for CWL Tool PR: #508
  • Config coveralls to measure code coverage diff on PR's PR: #526
  • Deprecate previous drafts PR: #520
  • Fix Makefile target for conformance code coverage Related issue: #334

Acknowledgement

I would like to thank a bunch of people for helping me throughout my journey. One of the best part of open source dev is to see other people willingly help each other out and collaborate to create amazing software. I'd like to thank my mentors Michael Crusoe, Anton Khodak and Janneke van der Zwaan. My interaction and collaboration with my mentors was great throughout. It was a big reason why I sticked well to the timeline and was able to keep up my progress. Special thanks to Michael. He was very responsive and helpful all along. He ended up giving useful insights throughout whenever I was stuck or something which I could have done better.

I would also like to thank Alexandr who helped me in mypy debugging and wrote a few stub files, Kapil who reviewed some of my code and I reviewed his, Petr for hanging around and giving ocassional advice. A big shout out to the entire CWL and mypy community: both of them were super active on Gitter and offered some very helpful advice on porting and mypy debugging.

I was able to sucessfully deliver my project objectives and thus, would consider it a successful summer work. I hope my work has postive impact on current and potential users of CWL Tool.

@tetron

This comment has been minimized.

Copy link

commented Aug 28, 2017

Thanks for all your great work Manvendra!

@manu-chroma

This comment has been minimized.

Copy link
Owner Author

commented Aug 31, 2017

Thank you, Peter. :)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
You can’t perform that action at this time.