v8.gyp cleanup for ARM cross-compiling

Cleanup v8.gyp.  This also drops a bunch of unused targets (the shell, d8, etc).

There are now two gyp variables to control whether we build snapshots and which target we are building for.  This allows you to easily cross-compile for arm and disable snapshots.

Continued in http://codereview.chromium.org/155218

[Joel Stanley (old), 2009-07-07 12:00:07]

Dean,

can you please take a look and see how it's progressing.  I might need a nudge
in the right direction; not quite sure what should be added next.

[Dean McNamee, 2009-07-08 13:08:02]

first pass

http://codereview.chromium.org/149243/diff/1006/6
File tools/gyp/v8.gyp (right):

http://codereview.chromium.org/149243/diff/1006/6#newcode380
Line 380: # v8 depends on v8_snapshot|v8_nosnapshot.
comment ins't very helpful

http://codereview.chromium.org/149243/diff/1006/6#newcode396
Line 396: # v8 depends on v8_snapshot|v8_nosnapshot.
comment isn't very helpful

http://codereview.chromium.org/149243/diff/1013/10#newcode1
Line 1: # v8.gyp, The Next Generation
can you restore the original header?

http://codereview.chromium.org/149243/diff/1013/10#newcode9
Line 9: 'build_snapshot%': 'true',
how about calling this v8_use_snapshot ?

http://codereview.chromium.org/149243/diff/1013/10#newcode107
Line 107: {
this isn't actually a library, make it makes sense to set the type to none ?

http://codereview.chromium.org/149243/diff/1013/10#newcode111
Line 111: # Subsequent targets use direct_dependant_settings to bring these
use the else of the conditional here.

http://codereview.chromium.org/149243/diff/1013/10#newcode123
Line 123: '../../src/allocation.h',
drop this comment

http://codereview.chromium.org/149243/diff/1013/10#newcode133
Line 133: '../../src/bootstrapper.h',
this isn't right, is it?

http://codereview.chromium.org/149243/diff/1013/10#newcode137
Line 137: '../../src/char-predicates-inl.h',
how about run_mksnapshot ?

http://codereview.chromium.org/149243/diff/1013/10#newcode149
Line 149: '../../src/compiler.cc',
drop this comment.

http://codereview.chromium.org/149243/diff/1013/10#newcode168
Line 168: '../../src/disassembler.h',
drop this comment, especially since the include doesn't propigate.  It's not
really common, it's that we're building a libv8_base.a.

http://codereview.chromium.org/149243/diff/1013/10#newcode420
Line 420: {
redo these platform- stuff like we talked about.

http://codereview.chromium.org/149243/diff/1013/10#newcode481
Line 481: '../../tools/js2c.py',
80 cols

[Joel Stanley (old), 2009-07-08 13:31:59]

There's one comment I wasn't sure about.

http://codereview.chromium.org/149243/diff/1013/10
File tools/gyp/v8.gyp (right):

http://codereview.chromium.org/149243/diff/1013/10#newcode1
Line 1: # v8.gyp, The Next Generation
On 2009/07/08 13:08:02, Dean McNamee wrote:
> can you restore the original header?

Done.

http://codereview.chromium.org/149243/diff/1013/10#newcode9
Line 9: 'v8_build_snapshot%': 'true',
On 2009/07/08 13:08:02, Dean McNamee wrote:
> how about calling this v8_use_snapshot ?

Done.

http://codereview.chromium.org/149243/diff/1013/10#newcode107
Line 107: 'type': '<(library)',
On 2009/07/08 13:08:02, Dean McNamee wrote:
> this isn't actually a library, make it makes sense to set the type to none ?

Done.

http://codereview.chromium.org/149243/diff/1013/10#newcode111
Line 111: },
On 2009/07/08 13:08:02, Dean McNamee wrote:
> use the else of the conditional here.

Done.

http://codereview.chromium.org/149243/diff/1013/10#newcode123
Line 123: # v8 depends on v8_snapshot|v8_nosnapshot.
On 2009/07/08 13:08:02, Dean McNamee wrote:
> drop this comment

Done.

http://codereview.chromium.org/149243/diff/1013/10#newcode133
Line 133: '../../src/snapshot.cc',
On 2009/07/08 13:08:02, Dean McNamee wrote:
> this isn't right, is it?

Hrm, where should it go?

http://codereview.chromium.org/149243/diff/1013/10#newcode137
Line 137: 'action_name': 'do_mksnapshot',
On 2009/07/08 13:08:02, Dean McNamee wrote:
> how about run_mksnapshot ?

Done.

http://codereview.chromium.org/149243/diff/1013/10#newcode149
Line 149: # v8 depends on v8_snapshot|v8_nosnapshot.
On 2009/07/08 13:08:02, Dean McNamee wrote:
> drop this comment.

Done.

http://codereview.chromium.org/149243/diff/1013/10#newcode168
Line 168: # Common includes and sources
On 2009/07/08 13:08:02, Dean McNamee wrote:
> drop this comment, especially since the include doesn't propigate.  It's not
> really common, it's that we're building a libv8_base.a.

Done.

http://codereview.chromium.org/149243/diff/1013/10#newcode420
Line 420: 'include', 'src/platform-posix.cc$'
On 2009/07/08 13:08:02, Dean McNamee wrote:
> redo these platform- stuff like we talked about.

Done.

http://codereview.chromium.org/149243/diff/1013/10#newcode481
Line 481: 'action': ['python', '../../tools/js2c.py', '<@(_outputs)', 'CORE',
'<@(library_files)'],
On 2009/07/08 13:08:02, Dean McNamee wrote:
> 80 cols

Done.

[Dean McNamee, 2009-07-08 13:39:16]

http://codereview.chromium.org/149243/diff/17/1015
File tools/gyp/v8.gyp (right):

http://codereview.chromium.org/149243/diff/17/1015#newcode130
Line 130: # This target is linked against by chrome.
I wouldn't say Chrome (although right now it's the only consumer).  I would just
say that this is the target that you probably want to depend on.  Or maybe just
drop the comment all together...

http://codereview.chromium.org/149243/diff/17/1015#newcode157
Line 157: '../../src/snapshot.cc',
shouldn't this be         '<(INTERMEDIATE_DIR)/snapshot.cc',

?

http://codereview.chromium.org/149243/diff/17/1015#newcode371
Line 371: # This needs magic when host is not ARM?
I would remove the comment

http://codereview.chromium.org/149243/diff/17/1015#newcode403
Line 403: # native ia32.
And this one too

http://codereview.chromium.org/149243/diff/17/1015#newcode432
Line 432: {
bring this { up and and do }] at the end, or do them all like this.  Right now
it's a mix.

http://codereview.chromium.org/149243/diff/17/1015#newcode470
Line 470: # Common target for both v8_snapshot and v8_nosnapshot.
don't see the help of this comment.

http://codereview.chromium.org/149243/diff/17/1015#newcode515
Line 515: # is unnecesssary but harmless.
yeah, can drop this comment.

[Joel Stanley (old), 2009-07-08 13:47:14]

http://codereview.chromium.org/149243/diff/17/1015
File tools/gyp/v8.gyp (right):

http://codereview.chromium.org/149243/diff/17/1015#newcode130
Line 130: # This target is linked against by chrome.
On 2009/07/08 13:39:16, Dean McNamee wrote:
> I wouldn't say Chrome (although right now it's the only consumer).  I would
just
> say that this is the target that you probably want to depend on.  Or maybe
just
> drop the comment all together...

Dropped.

http://codereview.chromium.org/149243/diff/17/1015#newcode157
Line 157: '../../src/snapshot.cc',
On 2009/07/08 13:39:16, Dean McNamee wrote:
> shouldn't this be         '<(INTERMEDIATE_DIR)/snapshot.cc',
> 
> ?

Ah, yes.  I was looking at the wrong line.

Done.

http://codereview.chromium.org/149243/diff/17/1015#newcode371
Line 371: # This needs magic when host is not ARM?
On 2009/07/08 13:39:16, Dean McNamee wrote:
> I would remove the comment

Done.

http://codereview.chromium.org/149243/diff/17/1015#newcode403
Line 403: # native ia32.
On 2009/07/08 13:39:16, Dean McNamee wrote:
> And this one too

Done.

http://codereview.chromium.org/149243/diff/17/1015#newcode432
Line 432: {
On 2009/07/08 13:39:16, Dean McNamee wrote:
> bring this { up and and do }] at the end, or do them all like this.  Right now
> it's a mix.

Done.

http://codereview.chromium.org/149243/diff/17/1015#newcode470
Line 470: # Common target for both v8_snapshot and v8_nosnapshot.
On 2009/07/08 13:39:16, Dean McNamee wrote:
> don't see the help of this comment.

Done.

http://codereview.chromium.org/149243/diff/17/1015#newcode515
Line 515: # is unnecesssary but harmless.
On 2009/07/08 13:39:16, Dean McNamee wrote:
> yeah, can drop this comment.

Done.

[Dean McNamee, 2009-07-08 13:55:47]

looking at the snapshotless build now.

http://codereview.chromium.org/149243/diff/1019/18
File tools/gyp/v8.gyp (right):

http://codereview.chromium.org/149243/diff/1019/18#newcode426
Line 426: ['OS=="linux"', {
this indention doesn't match mac/win

[Dean McNamee, 2009-07-08 14:04:02]

http://codereview.chromium.org/149243/diff/1022/20
File tools/gyp/v8.gyp (right):

http://codereview.chromium.org/149243/diff/1022/20#newcode437
Line 437: 'src/platform-posix.cc',
../../src

[Dean McNamee, 2009-07-08 14:42:31]

LGTM, and it builds on my machine with both snapshot and nosnapshot, and it
builds on ARM now too.

[Joel Stanley (old), 2009-07-08 14:43:02]

All issues should be addressed now.

I sucessfully cross-compiled and linked an ARM build with it.

http://codereview.chromium.org/149243/diff/1022/20
File tools/gyp/v8.gyp (right):

http://codereview.chromium.org/149243/diff/1022/20#newcode437
Line 437: 'src/platform-posix.cc',
On 2009/07/08 14:04:02, Dean McNamee wrote:
> ../../src

Done.

[Dean McNamee, 2009-07-08 14:47:00]

Kasper, sound ok?

[Kasper Lund, 2009-07-08 14:47:37]

I think you need to take care of the V8_NATIVE_REGEXP definition that was
recently added:

http://code.google.com/p/v8/source/diff?spec=svn2366&r=2366&format=side&path=...

It probably only makes sense for IA32 (not for ARM).