Absolute path fixes for gn.

Absolute path fixes for gn.

To avoid collisions object files and other build ouput is now prefixed
with ABS_PATH and the full path of the input file.
E.g. /some/path/foo.cc maps to //out/obj/ABS_PATH/some/path/foo.o

In addition there are several fixes for absolute paths on Windows:
The drive letter colon is skipped when parsing labels.
Transform C:\ to /C:\ where needed.

BUG=445454
NOPRESUBMIT=true

Committed: https://crrev.com/8808c5544ceb04ec281d40041af5272bdb7fad63
Cr-Commit-Position: refs/heads/master@{#315531}

[Fredrik Öhrn, 2015-01-20 13:37:14]

ohrn@opera.com changed reviewers:
+ brettw@chromium.org

[Fredrik Öhrn, 2015-01-20 13:42:11]

PTAL

[brettw, 2015-01-20 21:05:00]

https://codereview.chromium.org/857163002/diff/1/tools/gn/label.cc
File tools/gn/label.cc (right):

https://codereview.chromium.org/857163002/diff/1/tools/gn/label.cc#newcode98
tools/gn/label.cc:98: #if defined(OS_WIN)
On Windows, this accepts either "/C:/foo" or "C:/foo" I think. I'm not sure this
condition is correct. I think we should probably prohibit absolute paths that
don't begin in slashes here (the current behavior is to support paths returned
by third party tools, and these normally won't return label names). Can we also
add unit tests for these cases?

https://codereview.chromium.org/857163002/diff/1/tools/gn/label_pattern.cc
File tools/gn/label_pattern.cc (right):

https://codereview.chromium.org/857163002/diff/1/tools/gn/label_pattern.cc#ne...
tools/gn/label_pattern.cc:126: if (IsPathAbsolute(str)) {
Ditto

[Fredrik Öhrn, 2015-01-22 13:39:00]

On 2015/01/20 21:05:00, brettw wrote:
> https://codereview.chromium.org/857163002/diff/1/tools/gn/label.cc
> File tools/gn/label.cc (right):
> 
> https://codereview.chromium.org/857163002/diff/1/tools/gn/label.cc#newcode98
> tools/gn/label.cc:98: #if defined(OS_WIN)
> On Windows, this accepts either "/C:/foo" or "C:/foo" I think. I'm not sure
this
> condition is correct. I think we should probably prohibit absolute paths that
> don't begin in slashes here (the current behavior is to support paths returned
> by third party tools, and these normally won't return label names). Can we
also
> add unit tests for these cases?
> 

Done. We also fixed a couple of issues with absolute paths without a drive
letter
on Windows uncovered by the unit tests.

[Fredrik Öhrn, 2015-02-06 12:36:57]

Ping?

[brettw, 2015-02-06 16:36:32]

Oh, sorry I lost this. Will look soon.

[brettw, 2015-02-06 23:08:38]

LGTM. In the future, if I don't respond in <2 workdays it probably fell off my
list, so please don't feel like you need to wait so long to ping me.

https://codereview.chromium.org/857163002/diff/20001/tools/gn/label.cc
File tools/gn/label.cc (right):

https://codereview.chromium.org/857163002/diff/20001/tools/gn/label.cc#newcod...
tools/gn/label.cc:103: "Absolute paths must be on the form /C:\\");
"on the form" -> "of the form". Can you append to the long description

"but this is \"" + input + "\"."

(the reason is that the blame will often refer to a variable or an expression
and you won't know the value of the thing it's complaining about).

https://codereview.chromium.org/857163002/diff/20001/tools/gn/label_pattern.cc
File tools/gn/label_pattern.cc (right):

https://codereview.chromium.org/857163002/diff/20001/tools/gn/label_pattern.c...
tools/gn/label_pattern.cc:130: "Absolute paths must be on the form /C:\\");
Ditto from previous file.

[Fredrik Öhrn, 2015-02-10 09:49:20]

New patchsets have been uploaded after l-g-t-m from brettw@chromium.org

[Fredrik Öhrn, 2015-02-10 09:52:39]

The CQ bit was checked by ohrn@opera.com

[commit-bot: I haz the power, 2015-02-10 09:53:21]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/857163002/40001

[commit-bot: I haz the power, 2015-02-10 09:58:30]

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

[commit-bot: I haz the power, 2015-02-10 09:58:31]

Try jobs failed on following builders:
  chromium_presubmit on tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/chromium_presub...)

[Fredrik Öhrn, 2015-02-10 10:00:20]

On 2015/02/10 09:58:31, I haz the power (commit-bot) wrote:
> Try jobs failed on following builders:
>   chromium_presubmit on tryserver.chromium.linux (JOB_FAILED,
>
http://build.chromium.org/p/tryserver.chromium.linux/builders/chromium_presub...)

Seems I can't commit this myself.

[Mostyn Bramley-Moore, 2015-02-10 10:04:51]

On 2015/02/10 10:00:20, Fredrik Öhrn wrote:
> On 2015/02/10 09:58:31, I haz the power (commit-bot) wrote:
> > Try jobs failed on following builders:
> >   chromium_presubmit on tryserver.chromium.linux (JOB_FAILED,
> >
>
http://build.chromium.org/p/tryserver.chromium.linux/builders/chromium_presub...)
> 
> Seems I can't commit this myself.

Try adding "NOPRESUBMIT=true" to the bug description.

[Fredrik Öhrn, 2015-02-10 10:09:33]

The CQ bit was checked by ohrn@opera.com

[commit-bot: I haz the power, 2015-02-10 10:10:26]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/857163002/40001

[commit-bot: I haz the power, 2015-02-10 10:18:46]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-02-10 10:19:21]

Patchset 3 (id:??) landed as
https://crrev.com/8808c5544ceb04ec281d40041af5272bdb7fad63
Cr-Commit-Position: refs/heads/master@{#315531}

[Fredrik Öhrn, 2015-02-10 10:21:56]

On 2015/02/10 10:04:51, Mostyn Bramley-Moore wrote:
> On 2015/02/10 10:00:20, Fredrik Öhrn wrote:
> > Seems I can't commit this myself.
> 
> Try adding "NOPRESUBMIT=true" to the bug description.

Thanks for the tip.