win ninja: include .pdb in linker outputs

win ninja: include .pdb in linker outputs

This is "just more correct" (will e.g. make -t clean work a bit
better) and also allows syzygy rules that process .pdbs to work.

BUG=chromium:325815
R=thakis@chromium.org

Committed: https://code.google.com/p/gyp/source/detail?r=1832

[Nico, 2014-01-08 21:44:13]

https://codereview.chromium.org/126443004/diff/150001/pylib/gyp/generator/nin...
File pylib/gyp/generator/ninja.py (right):

https://codereview.chromium.org/126443004/diff/150001/pylib/gyp/generator/nin...
pylib/gyp/generator/ninja.py:1114: pdbname = self.msvs_settings.GetPDBName(
# Must match /PDB: flag in the 'link' rule.

https://codereview.chromium.org/126443004/diff/150001/pylib/gyp/generator/nin...
pylib/gyp/generator/ninja.py:1115: config_name, self.ExpandSpecial, output +
'.pdb')
default=output + '.pdb'

https://codereview.chromium.org/126443004/diff/150001/pylib/gyp/generator/nin...
pylib/gyp/generator/ninja.py:1623: # overriding default settings earlier in the
command line.
# The /PDB: argument must match the pdbname written by WriteLinkForArch.

https://codereview.chromium.org/126443004/diff/150001/pylib/gyp/generator/nin...
pylib/gyp/generator/ninja.py:2130: if False: #params['parallel']:
I assume you don't mean to land this part :-)

https://codereview.chromium.org/126443004/diff/150001/pylib/gyp/msvs_emulatio...
File pylib/gyp/msvs_emulation.py (right):

https://codereview.chromium.org/126443004/diff/150001/pylib/gyp/msvs_emulatio...
pylib/gyp/msvs_emulation.py:329: return
expand_special(self.ConvertVSMacros(output_file, config=config))
This won't match the /PDB: flag passed to the linker, right? Is that not a
problem?

https://codereview.chromium.org/126443004/diff/150001/test/win/gyptest-link-p...
File test/win/gyptest-link-pdb-output.py (right):

https://codereview.chromium.org/126443004/diff/150001/test/win/gyptest-link-p...
test/win/gyptest-link-pdb-output.py:17: test =
TestGyp.TestGyp(formats=['ninja'])
Does this not pass with msvs?

[scottmg, 2014-01-08 21:54:56]

https://codereview.chromium.org/126443004/diff/150001/pylib/gyp/generator/nin...
File pylib/gyp/generator/ninja.py (right):

https://codereview.chromium.org/126443004/diff/150001/pylib/gyp/generator/nin...
pylib/gyp/generator/ninja.py:1114: pdbname = self.msvs_settings.GetPDBName(
On 2014/01/08 21:44:13, Nico wrote:
> # Must match /PDB: flag in the 'link' rule.

It's confusing, but it actually doesn't have to. The link rule sets some name,
and then GetPDBName as used by GetLdflags overrides it with its own setting for
/PDB.

...

Hm, actually, maybe I should just remove it entirely from the rule. I'll look
into that.

https://codereview.chromium.org/126443004/diff/150001/pylib/gyp/generator/nin...
pylib/gyp/generator/ninja.py:1115: config_name, self.ExpandSpecial, output +
'.pdb')
On 2014/01/08 21:44:13, Nico wrote:
> default=output + '.pdb'

Done. (and above)

https://codereview.chromium.org/126443004/diff/150001/pylib/gyp/generator/nin...
pylib/gyp/generator/ninja.py:2130: if False: #params['parallel']:
On 2014/01/08 21:44:13, Nico wrote:
> I assume you don't mean to land this part :-)

Ahem. I need to look into why the stack traces are wrong in parallel in the
generator. :/

https://codereview.chromium.org/126443004/diff/150001/test/win/gyptest-link-p...
File test/win/gyptest-link-pdb-output.py (right):

https://codereview.chromium.org/126443004/diff/150001/test/win/gyptest-link-p...
test/win/gyptest-link-pdb-output.py:17: test =
TestGyp.TestGyp(formats=['ninja'])
On 2014/01/08 21:44:13, Nico wrote:
> Does this not pass with msvs?

No, because you can't ask it to build a file (the pdb), only a project (== gyp
target).

[Nico, 2014-01-08 22:06:46]

Ok, blind lgtm :-)

(I think you might not have uploaded a new patch set. Or if you want to look
into the /PDB: flag thing in this CL, tell me if you want me to take another
look)

[scottmg, 2014-01-08 22:49:48]

Sorry, was looking at /PDB. It's removed from the rule now. PTanotherL.

[Nico, 2014-01-08 23:00:36]

lgtm

[scottmg, 2014-01-08 23:08:16]

Committed patchset #6 manually as r1832 (presubmit successful).