1. 23 May, 2019 1 commit
  2. 13 May, 2019 2 commits
  3. 07 May, 2019 1 commit
    • Harald Sitter's avatar
      prevent generic details from factorizing · 57c8e807
      Harald Sitter authored
      with latest master I am seeing tests (supposedly) failing as you would
      expect, but then the output json contains generic details with no
      worthwhile information. prevent such details from factorizing.
      regardless of other defects that may play into this result there is not
      a single reason where factorizing a worthless detail is something we need
      or want to do. instead ignore them entirely. this is a bit shoddy mapping
      but TBH the detail entry in of itself is too
  4. 06 May, 2019 2 commits
  5. 15 Jun, 2018 1 commit
  6. 14 Jun, 2018 1 commit
  7. 02 May, 2018 1 commit
  8. 27 Apr, 2018 1 commit
  9. 24 Apr, 2018 3 commits
  10. 18 Apr, 2018 2 commits
  11. 13 Feb, 2018 1 commit
    • Harald Sitter's avatar
      archive the wok relative · e6ae0008
      Harald Sitter authored
      previously we'd have to dig into wok to get to the wok's artifacts, since
      they are the primary artifacts that is fairly daft though. by diring
      into the wok first we can archive them into the root of the artifacts dir
  12. 12 Feb, 2018 2 commits
  13. 08 Feb, 2018 4 commits
    • Harald Sitter's avatar
    • Harald Sitter's avatar
      add an approximation fallback · c994af63
      Harald Sitter authored
      if >1 wants the data
      but =0 need the data
      then fall back to the closest approximation and hope the junit printer code
      for the approximation can properly handle the approximate shape.
      this is necessary to handle screenshots with no needle. this is different
      from a regular screenshot that wasn't meant to match, it is caused
      by assert_screen(tags) without any needles matching tags.
      this blob e.g.
      {:result=>"unk", :tags=>["ubuntu-upgrade-fetcher"], :screenshot=>"upgrade_bionic-29.png", :frametime=>["10.71", "10.75"]}
      which is basically a NeedleDetail but without the actual needles field,
      since it had no needles.
      we can represent this easily enough in a NeedleDetail if we simply init
      This could have been represented using a dedicated class BUT, it's easy
      enough to do this way.
      Should this became a problem in the future we can always unpack this
      by simply adding a class that fully matches the blob (match > approx).
      However to keep the code dry we'd have to make the class inheritance of
      the details more complex, so I am shying away from this for now.
      (a NeedleMatchDetail is a NoNeedleMatchDetail AND a NeedleDetail
       a NeedleDetail is a NoNeedleDetail - that's diamond inheritance which
       Ruby doesn't like, so representing it this way is a world of trouble)
    • Harald Sitter's avatar
      init with more sane types · ad4b6509
      Harald Sitter authored
    • Harald Sitter's avatar
      give output when we failed to map a blob · 0adf9694
      Harald Sitter authored
  14. 15 Jan, 2018 1 commit
    • Harald Sitter's avatar
      refine handling of chain shots · 85a7816c
      Harald Sitter authored
      see comments all over
      basically multiple screenshots may contribute to the same check but openqa
      doesn't all mark them as the result of the final outcome of the check.
      so, we do this manually after the fact by running through all details,
      finding chains of unknown results and finalizing them using results of
      a subsequent final detail
  15. 10 Jan, 2018 1 commit
  16. 19 Dec, 2017 1 commit
  17. 14 Dec, 2017 2 commits
    • Harald Sitter's avatar
      revise soft failure handling · 405c764e
      Harald Sitter authored
      we now use a different class to represent soft failures. this way we
      can easily render them separately from text when converting to junit
    • Harald Sitter's avatar
      redo the way we parse results · 3bf446ce
      Harald Sitter authored
      now properly modeling the json from osautoinst which makes it easier to
      write tests around this in the future and classify the entire thing
      so we can write case representations of the invidiual types
  18. 05 Dec, 2017 1 commit
    • Harald Sitter's avatar
      rejigger junit a bit · ada4f12c
      Harald Sitter authored
      - instead of globbing hold on to test_order.json to find run tests
      - inject index into name of the suites as per test_order to keep them
        ordered in jenkins as well, otherwise they'll be shown in random(?) order
        which sucks given they are very much ordered and order may matter WRT
        why a test failed
      - junit.rb now raises if any test had any failed case, this is to
        workaround isotovideo always coming back 0 even if it had errors, for
        jenkins we need to reflect errors more aggressively
  19. 01 Dec, 2017 1 commit
  20. 28 Nov, 2017 1 commit
  21. 29 Jun, 2017 2 commits
  22. 21 Jun, 2017 7 commits
  23. 14 Jun, 2017 1 commit