Improvements to build-script.py #521
Conversation
|
@swift-ci Please test |
No functionality change
This guarantees that we fail execution of build-script.py if a subcommand fails.
ahoppen
force-pushed
the
pr/build-script-improvements
branch
from
e52e96f to
de3c8f8
Compare
This reduces the time to gyb-generate sources from ~12s to ~4s.
ahoppen
force-pushed
the
pr/build-script-improvements
branch
from
de3c8f8 to
2d197b2
Compare
|
I also added a new improvement now to run gyb-generation on multiple threads, speeding it up by 3x from ~12s to ~4s: de3c8f8 |
|
@swift-ci Please test |
| else: | ||
| lit_success = run_lit_tests( | ||
| if not skip_lit_tests: | ||
| run_lit_tests( |
There was a problem hiding this comment.
I know this is the current behavior but why we don't run xctest cases when lit-tests fails? I feel we should run them both unconditionally, and raise a failure when either of them fails.
There was a problem hiding this comment.
I think there are three reasons:
- I like the way build-script fails early. That way the log output of lit-based tests are closer to the bottom (and thus easier to read) than when we continue to execute xctest commands
- It follows the current paradigm of the build-script that if one command fails it shouldn’t execute any other commands
- It’s slightly easier to implement this way
If you have strong opinions on this, we can change it but I like it the way it is.
There was a problem hiding this comment.
I see. I agree that it's nice we can see the failures closer to the bottom. I still think it's also nice too all failures for both test suites, but it's not a strong opinion, and I can skip lit tests if I want. So let's keep it as it is.
|
@swift-ci Please test |
check_callfor subprocesses to make sure build-script.py fails with a non-zero exit code if any subprocess fails.