Fix Linux coverage which doesn't like a gyp 'inputs' to contain non-files....

Fix Linux coverage which doesn't like a gyp 'inputs' to contain non-files.
Tested on Linux and Mac.


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=48255

[viettrungluu, 2010-05-26 04:14:50]

http://codereview.chromium.org/2262001/diff/2001/3001
File chrome/chrome_tests.gypi (right):

http://codereview.chromium.org/2262001/diff/2001/3001#newcode2387
chrome/chrome_tests.gypi:2387: '../ipc/ipc.gyp:ipc_tests',
You broke the indentation here.

http://codereview.chromium.org/2262001/diff/2001/3001#newcode2424
chrome/chrome_tests.gypi:2424: # Technically I want inputs to be the list of
Could you label this with TODO(jrg), so we know who "I" is (and who to ask when
there are any questions)?

http://codereview.chromium.org/2262001/diff/2001/3001#newcode2429
chrome/chrome_tests.gypi:2429: 'inputs': [ 'chrome/chrome_tests.gypi' ],
Will this really make this run? (It seems to me that it would only be run each
time chrome_tests.gypi is touched.)

If the objective is to make it run, could you give it a nonexistent file as
'outputs'? (The documentation says: "If any output is missing or is outdated
relative to any input, the custom action will be invoked.")

[John Grabowski, 2010-05-26 04:20:34]

http://codereview.chromium.org/2262001/diff/2001/3001
File chrome/chrome_tests.gypi (right):

http://codereview.chromium.org/2262001/diff/2001/3001#newcode2387
chrome/chrome_tests.gypi:2387: '../ipc/ipc.gyp:ipc_tests',
On 2010/05/26 04:14:50, viettrungluu wrote:
> You broke the indentation here.

fixed

http://codereview.chromium.org/2262001/diff/2001/3001#newcode2424
chrome/chrome_tests.gypi:2424: # Technically I want inputs to be the list of
On 2010/05/26 04:14:50, viettrungluu wrote:
> Could you label this with TODO(jrg), so we know who "I" is (and who to ask
when
> there are any questions)?

done

http://codereview.chromium.org/2262001/diff/2001/3001#newcode2429
chrome/chrome_tests.gypi:2429: 'inputs': [ 'chrome/chrome_tests.gypi' ],
On 2010/05/26 04:14:50, viettrungluu wrote:
> Will this really make this run? (It seems to me that it would only be run each
> time chrome_tests.gypi is touched.)
> 
> If the objective is to make it run, could you give it a nonexistent file as
> 'outputs'? (The documentation says: "If any output is missing or is outdated
> relative to any input, the custom action will be invoked.")

Nope.  For Linux/Make, a non-existent file as input causes a make error.  That
is the reason why Linux coverage is broken right now.

[viettrungluu, 2010-05-26 04:31:44]

Okay, LGTM with a bit more explanation in the comments. In any case, running
coverage is better than not running coverage.

On 2010/05/26 04:20:34, John Grabowski wrote:
> http://codereview.chromium.org/2262001/diff/2001/3001
> File chrome/chrome_tests.gypi (right):
> 
> http://codereview.chromium.org/2262001/diff/2001/3001#newcode2387
> chrome/chrome_tests.gypi:2387: '../ipc/ipc.gyp:ipc_tests',
> On 2010/05/26 04:14:50, viettrungluu wrote:
> > You broke the indentation here.
> 
> fixed
> 
> http://codereview.chromium.org/2262001/diff/2001/3001#newcode2424
> chrome/chrome_tests.gypi:2424: # Technically I want inputs to be the list of
> On 2010/05/26 04:14:50, viettrungluu wrote:
> > Could you label this with TODO(jrg), so we know who "I" is (and who to ask
> when
> > there are any questions)?
> 
> done
> 
> http://codereview.chromium.org/2262001/diff/2001/3001#newcode2429
> chrome/chrome_tests.gypi:2429: 'inputs': [ 'chrome/chrome_tests.gypi' ],
> On 2010/05/26 04:14:50, viettrungluu wrote:
> > Will this really make this run? (It seems to me that it would only be run
each
> > time chrome_tests.gypi is touched.)
> > 
> > If the objective is to make it run, could you give it a nonexistent file as
> > 'outputs'? (The documentation says: "If any output is missing or is outdated
> > relative to any input, the custom action will be invoked.")
> 
> Nope.  For Linux/Make, a non-existent file as input causes a make error.  That
> is the reason why Linux coverage is broken right now.