[swarming] use client path from recipe_engine/swarming

We cannot always assume that there is swarming client in the specific directory of repository.
e.g. v8's tester builder, ios's internal repository
Instead let me take swarming client always from CIPD.

To access recipe_engine/swarming inside build/swarming, this CL uses build/swarming_client as wrapper for on_path().

This is an important change: this now ignores the pinned version in the project's repository
and instead always use the pinned version in the recipe but only for task result collection.

Triggering is not affected.

Bug: 894045
Recipe-Nontrivial-Roll: build_limited_scripts_slave
Change-Id: I4105b74738a64a280422f99bad472e1f7597e32f
Reviewed-on: https://chromium-review.googlesource.com/c/1378754
Commit-Queue: Takuto Ikuta <tikuta@chromium.org>
Reviewed-by: Marc-Antoine Ruel <maruel@chromium.org>
242 files changed
tree: 131449407d49c04c87e60956312e2a9686bb122b
  1. infra/
  2. masters/
  3. scripts/
  4. site_config/
  5. slave/
  6. tests/
  7. third_party/
  8. .gitattributes
  9. .gitignore
  10. .vpython
  11. codereview.settings
  12. DEPS
  13. environment.cfg.py
  14. LICENSE
  15. OWNERS
  16. PRESUBMIT.py
  17. README.md
  18. WATCHLISTS
README.md

build

Hi build contributor! If you do any change in scripts/master/ or touching any master's html/ directories, you must restart master.chromium.fyi first and ensure that it still works before restarting other masters.

Recipes

If you're here to make a change to ‘recipes’ (the code located in scripts/slave/recipes*), please take a look at the README for more information pertaining to recipes.