TODO

 - use test verbosity to also add more details to existing issues but always
   one level less verbose than for new tests

 - check install error when cloned from gitlab w/o tags maybe based on pylibrary cookiecutter template? missing tag is causing version to fail
  - tags added but should also work without, try "bumpversion"? see cookiecutter-pylibrary template

 - offer to automatically post comment on webpage

 - for new issues,
  - offer hints what to do, e.g 'fix the needle', etc.
  - parse and show link (clickable) for browser OR auto-open in browser
  - show detailed fail reason (failed needles are always shown, easy)

 - for fixed issues,
  - also offer link to crosscheck
  - offer to close bug


 - compare architectures against each other, if same test fails for all
   architectures, report as "common issues"


 - consider "build number drift", e.g. some builds have
   i586 and x86_64 and some the corresponding 'arm' tests

 - for further investigation of failed issues
  - compare list of needle candidates for failed needles
   - empty list
   - less needles
   - more needles
   - changed names


 - search for open bugs in bugzilla

  - how to find bugs which only have a test_url which does not specify the
    test module? e.g. if someone only mentions
    https://openqa.opensuse.org/tests/167348/file/autoinst-log.txt


  - diff os-autoinst.txt logfiles, e.g. download and offer wizard like in
    desired user diff editor
   - can skip differing timestamps and near matches, maybe python difflib can
     do this?


  - offer links to
   - autoinst.txt (should be easy, it is the test-url + '/file/autoinst-log.txt', e.g. https://openqa.opensuse.org/tests/167348/file/autoinst-log.txt')
   - the test source code
   - the git changelog of test + openQA + os-autoinst
   - the OBS packages changed during period


 - redirect report to file
 - offer to open urls to failed tests in browser
 - offer to open bugs in bugzilla or just text template (could enter new bugs
   but don't submit yet for review), e.g. based on
   https://bugzilla.suse.com/show_bug.cgi?id=959694
 - template for failed openqa product bugs in wiki
 - improve detection about "new product issue" or "new openqa issue".
   Assumption: If the source code of the test did not change
   (previous->current) AND the needle candidates (or tags) did not change (AND
   maybe the test one step in before also did not change and not fail) THEN a
   failed test is a new product issue
 - compare log output, e.g. see
  - http://stackoverflow.com/questions/6486450/python-compute-list-difference
  - http://stackoverflow.com/questions/3462143/get-difference-between-two-lists
  - https://pypi.python.org/pypi/fuzzywuzzy
  - http://streamhacker.com/2011/10/31/fuzzy-string-matching-python/
  - http://chairnerd.seatgeek.com/fuzzywuzzy-fuzzy-string-matching-in-python/
  - http://marcobonzanini.com/2015/02/25/fuzzy-string-matching-in-python/
    - http://stackoverflow.com/questions/2923420/what-is-a-simple-fuzzy-string-matching-algorithm-in-python


 - package for openSUSE distributions
  - make open soure
  - use osc to create OBS project
  - build spec file with "python setup.py bdist_rpm --force-arch x86_64"
  - build with "osc build"
  - upload
  - add SR