1. 23 Jun, 2020 1 commit
    • Harald Sitter's avatar
      write to new path · 72f7c795
      Harald Sitter authored
      paths were rejiggered in 55ee906d486a22e3b97dbef172cb258b1ff287a7 of
      ci-tooling repo
      72f7c795
  2. 03 Jun, 2020 4 commits
    • Harald Sitter's avatar
      GIT_SILENT style++ · b31ea355
      Harald Sitter authored
      b31ea355
    • Harald Sitter's avatar
      s/anongit/invent · 753f850b
      Harald Sitter authored
      anongit is going away
      753f850b
    • Harald Sitter's avatar
      compatibility bandaid for project->product data mapping · ca19066d
      Harald Sitter authored
      project paths in the API are now invent paths, but product data
      in ci-tooling still talks about the legacy project paths. this caused
      a divide in what metadata is talking about resultin in entirely broken
      extraction as it didn't even manage to resolve plasma-workspace to
      an artifact for download.
      
      to resolve this for now we'll fake our internal representation of the api
      blobs to still be path as per the legacy paths rather than the new ones.
      
      simply put while plasma-workspace is now actually `plasma/plasma-workspace`
      we'll act as though it was still `kde-workspace/plasma-workspace`.
      this is based on a temporary legacy attribute in the API that is bound
      to go away at some point. when that happens ci-tooling will also have
      moved to the new paths though, so we can just remove the override of
      the id again
      
      this requires batch resolution of all projects to get to the legacy paths
      unfortunately. to keep speed impact low we'll aggressively thread this
      (resolution would ordinarily be done lazily in a promise anyway, so
       over the entire run this shouldn't make a difference, if anything its
       probably slightly faster now because previously the .get() was bound
       to the execution pool at 8 threads while now we can use an extraordinary
       amount of threads because we basically do no work other than the https
       requests)
      
      I've also rejiggered the override a code and refactored the methods a bit
      because they were fairly horrible to read
      ca19066d
    • Harald Sitter's avatar
      raise more readable errors when base artifact retrieval fails · 2cb1a4bc
      Harald Sitter authored
      previously this would raise without a message, which is reasonable enough
      as one can simply check out the code. but since this has happend now I
      feel like a more digestible string is in order
      2cb1a4bc
  3. 05 May, 2020 1 commit
  4. 01 May, 2020 1 commit
  5. 23 Jan, 2020 2 commits
  6. 02 Nov, 2019 1 commit
  7. 18 Oct, 2019 2 commits
  8. 08 Oct, 2019 2 commits
  9. 22 Aug, 2019 2 commits
  10. 21 Aug, 2019 2 commits
  11. 06 Aug, 2019 1 commit
  12. 05 Aug, 2019 3 commits
  13. 04 Aug, 2019 1 commit
  14. 03 Aug, 2019 2 commits
  15. 02 Aug, 2019 3 commits
  16. 01 Aug, 2019 2 commits
  17. 30 Jul, 2019 2 commits
  18. 19 Jul, 2019 1 commit
  19. 19 Jun, 2019 1 commit
  20. 05 Jun, 2019 1 commit
  21. 04 Jun, 2019 2 commits
  22. 02 Jun, 2019 3 commits