Add support for create_installer_archive.py to generate a depfile

Add support for create_installer_archive.py to generate a depfile

Adds functionality to create_installer_archive.py to generate a .d file
that includes all of the files that it packages into chrome[.packed].7z.
This is done by tracking the source of all the files that are copied
into the staging dir that is given to 7za to package the .7z file.

The depfile is then used in a gyp 'depfile' to correctly specify the
implicit dependencies for the mini_installer target. Previously it used
a fake output of 'xxx2.out' to cause it to rebuild every time (which is
what we're trying to avoid in this change).

Additionally, switch from a rule to an action in gyp -- gyp currently
only supports 'depfile' on actions, and as there's only one file being
built here, they're equivalent.

BUG=342974, 451499
Committed: https://crrev.com/f79c2e881fb94775cdac5679795ae3e6957fc487
Cr-Commit-Position: refs/heads/master@{#314015}

[scottmg, 2015-01-29 23:32:41]

https://codereview.chromium.org/887673003/diff/1/chrome/installer/mini_instal...
File chrome/installer/mini_installer.gyp (left):

https://codereview.chromium.org/887673003/diff/1/chrome/installer/mini_instal...
chrome/installer/mini_installer.gyp:195:
'<(PRODUCT_DIR)/<(RULE_INPUT_NAME).packed.7z',
this expanded to "chrome.release.packed.7z" which was never generated, i assume
accidentally.

https://codereview.chromium.org/887673003/diff/1/chrome/installer/mini_instal...
chrome/installer/mini_installer.gyp:246: 'xxx2.out',
this was to force an always-build which we're trying to avoid

https://codereview.chromium.org/887673003/diff/1/chrome/installer/mini_instal...
chrome/installer/mini_installer.gyp:247: '<(PRODUCT_DIR)/<(RULE_INPUT_NAME).7z',
same as .packed.7z above

https://codereview.chromium.org/887673003/diff/1/chrome/installer/mini_instal...
File chrome/installer/mini_installer.gyp (right):

https://codereview.chromium.org/887673003/diff/1/chrome/installer/mini_instal...
chrome/installer/mini_installer.gyp:251:
'<(INTERMEDIATE_DIR)/installer_archive.d',
this doesn't work yet (gyp only supports it on actions, not rules)

[scottmg, 2015-01-30 00:07:14]

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

[scottmg, 2015-01-30 00:08:45]

grt: installer-y bits + OWNERS
thakis: knowing all about depfile/ninja.

[grt (UTC plus 2), 2015-01-30 19:03:27]

This is awesome. Thanks for tackling it. The python looks better than what I
would have written, but consider getting someone with python readability to
review if that's a concern for you. LGTM.

[scottmg, 2015-01-30 19:22:25]

The CQ bit was checked by scottmg@chromium.org

[commit-bot: I haz the power, 2015-01-30 19:23:43]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/887673003/40001

[commit-bot: I haz the power, 2015-01-30 20:12:37]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-01-30 20:12:37]

Try jobs failed on following builders:
  win_chromium_compile_dbg_ng on tryserver.chromium.win
(http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_comp...)

[scottmg, 2015-01-30 21:07:09]

On 2015/01/30 20:12:37, I haz the power (commit-bot) wrote:
> Try jobs failed on following builders:
>   win_chromium_compile_dbg_ng on tryserver.chromium.win
>
(http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_comp...)

Shucks, that's a real failure -- Just testing on a component debug build, but
it's probably because it's copying the runtime dlls from C to a build dir on E I
guess.

Only the output in the depfile has to be relative-ized (so that it matches the
name in the build.ninja files), so I'll switch to doing only that.

[scottmg, 2015-01-30 21:52:37]

Patchset #4 (id:60001) has been deleted

[scottmg, 2015-01-30 21:52:42]

Patchset #3 (id:40001) has been deleted

[scottmg, 2015-01-30 21:54:52]

Now doesn't relpath the inputs so it can work across drives, fixes a bug in the
component dll copy, and escapes spaces in the paths when they're copied from
"c:\program files..." for the dlls copied component mode.

ptaskim

[Nico, 2015-01-30 22:51:14]

lgtm

Does that script have tests, perchance? :-)

[scottmg, 2015-01-30 22:52:37]

The CQ bit was checked by scottmg@chromium.org

[scottmg, 2015-01-30 22:53:08]

On 2015/01/30 22:51:14, Nico wrote:
> lgtm
> 
> Does that script have tests, perchance? :-)

We have ... the Official builder that a TPM will yell about shortly?

[commit-bot: I haz the power, 2015-01-30 22:53:37]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/887673003/80001

[commit-bot: I haz the power, 2015-01-30 22:57:12]

Committed patchset #3 (id:80001)

[commit-bot: I haz the power, 2015-01-30 22:57:58]

Patchset 3 (id:??) landed as
https://crrev.com/f79c2e881fb94775cdac5679795ae3e6957fc487
Cr-Commit-Position: refs/heads/master@{#314015}

[grt (UTC plus 2), 2015-01-30 23:13:37]

On 2015/01/30 22:51:14, Nico wrote:
> lgtm
> 
> Does that script have tests, perchance? :-)

In a sense. test_installer.py runs the installer, and will give you a red tryrun
if it doesn't seem to install something resembling Chromium/Chrome.

[scottmg, 2015-01-30 23:16:33]

On 2015/01/30 23:13:37, grt wrote:
> On 2015/01/30 22:51:14, Nico wrote:
> > lgtm
> > 
> > Does that script have tests, perchance? :-)
> 
> In a sense. test_installer.py runs the installer, and will give you a red
tryrun
> if it doesn't seem to install something resembling Chromium/Chrome.

Oops, did I run that in the default set above?

[grt (UTC plus 2), 2015-01-31 02:20:26]

On 2015/01/30 23:16:33, scottmg wrote:
> On 2015/01/30 23:13:37, grt wrote:
> > On 2015/01/30 22:51:14, Nico wrote:
> > > lgtm
> > > 
> > > Does that script have tests, perchance? :-)
> > 
> > In a sense. test_installer.py runs the installer, and will give you a red
> tryrun
> > if it doesn't seem to install something resembling Chromium/Chrome.
> 
> Oops, did I run that in the default set above?

For a static build, yes:
http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_rel_....

For a component build, possibly not. I think the dbg win bots would run it.