ninja: Correctly handle copies steps in sourceless shared_library targets.

ninja: Correctly handle copies steps in sourceless shared_library targets.

The other generators I tested (make, xcode) all get this right already.

BUG=chromium:552692
R=sbaig1@bloomberg.net, scottmg@chromium.org

Committed: https://chromium.googlesource.com/external/gyp/+/b85ad3e578da830377dbc1843aa4fbc5af17a192

[Nico, 2015-12-12 19:51:31]

thakis@chromium.org changed reviewers:
+ scottmg@chromium.org

[Nico, 2015-12-12 19:51:32]

`time python gyptest.py  -f ninja` takes a _lot_ longer to run that I remembered
(but I haven't run it in a long time and maybe it's just better in my memory)

[Nico, 2015-12-14 19:11:24]

thakis@chromium.org changed reviewers:
+ sbaig1@bloomberg.net

[Nico, 2015-12-14 19:11:25]

Shezan, can you review this maybe? Scott is out sick, and you already helped me
with the test :-)

If you look at WriteTarget(), you'll see that all the branches that don't call
WriteLink() pass compile_deps as order-only deps. WriteLink() was missing that,
this CL adds it there (needed in case there are no compile steps that can depend
on them).

[scottmg, 2015-12-14 19:24:52]

lgtm

https://codereview.chromium.org/1518313002/diff/80001/test/copies/gyptest-sou...
File test/copies/gyptest-sourceless-shared-lib.py (right):

https://codereview.chromium.org/1518313002/diff/80001/test/copies/gyptest-sou...
test/copies/gyptest-sourceless-shared-lib.py:15: chdir='src')
It looks like this would fit on the line above.

[Shezan Baig (Bloomberg), 2015-12-14 19:26:11]

LGTM

But instead of having to specify TargetMachine in the gyp file, we could
probably make msvs_emulation.py automatically add a /MACHINE flag if it detects
that there are no obj files.  Not sure if it is worth it though, especially
since most people would have TargetMachine in their common.gypi files anyway.

[Nico, 2015-12-14 19:53:40]

Description was changed from

==========
ninja: Correctly handle copies steps in sourceless shared_library targets.

The other generators I tested (make, xcode) all get this right already.

BUG=chromium:552692
==========

to

==========
ninja: Correctly handle copies steps in sourceless shared_library targets.

The other generators I tested (make, xcode) all get this right already.

BUG=chromium:552692
R=sbaig1@bloomberg.net, scottmg@chromium.org

Committed:
https://chromium.googlesource.com/external/gyp/+/b85ad3e578da830377dbc1843aa4...
==========

[Nico, 2015-12-14 19:53:41]

Committed patchset #6 (id:100001) manually as
b85ad3e578da830377dbc1843aa4fbc5af17a192 (presubmit successful).

[Nico, 2015-12-14 19:54:53]

Thanks!

We can make msvs_emulation smarter/more complicated if another 2 tests need this
too :-)

https://codereview.chromium.org/1518313002/diff/80001/test/copies/gyptest-sou...
File test/copies/gyptest-sourceless-shared-lib.py (right):

https://codereview.chromium.org/1518313002/diff/80001/test/copies/gyptest-sou...
test/copies/gyptest-sourceless-shared-lib.py:15: chdir='src')
On 2015/12/14 19:24:51, scottmg (out) wrote:
> It looks like this would fit on the line above.

Done.