On 2013/10/29 14:02:54, rmcilroy wrote:
I realize this will increase test time by about 25% - that's why I didn't
introduce it in the original CL - but I think it's important to test with
this
flag enabled since it significantly changes generated prologue code (and
will
change other generated code as more optimizations are added), and the
run-tests.py tests found some issues which weren't apparent when running
the
JS
benchmarks.
You can test this locally via e.g.
make -j32 arm.check TESTFLAGS="--extra-flags --optimize-for-size"
If we kept every new flag for every new feature as a variant in the test
suite,
we effectively be dead by now. :-/
Having a runner (for each arch) with this flag enabled would be OK, but
are
there extra runners available on which we can enable this flag (from what
I
can
see, there is only a single ARM runner on the V8 waterfall, and I think we
want
to test both with and without the flag). Thoughts?
What we normally do is that some runners have some flags enabled, others
don't.
This way we don't really have a testing *matrix*, but at least some
coverage.
I know that this is annoying, and of course everybody likes to get good test
coverage for his own new feature, but it is important to keep test times
manageable. Recently there was a very visible tendency that people obviously
didn't run the test suite at all before committing, and we should actively
fight
that tendency...
https://codereview.chromium.org/49633004/
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.
