Rename target-specific binaries built for host.

Rename target-specific binaries built for host.

Rename v8_base, v8_nosnapshot, and mksnapshot to include a suffix of
.<(v8_target_arch). This allows multiple target architectures to be
built in the Android build system, which uses a single shared directory
structure for building host binaries.

BUG=

Committed: http://code.google.com/p/v8/source/detail?r=14209

[Torne, 2013-03-21 11:49:03]

Hi Jakob,

As we discussed in email, here's a patch that implements the renamings required
to make building multiple architectures work in the Android build system. Only
three targets are renamed. There is a fourth target built for host (js2c) but
its makefiles do not contain any architecture dependency and thus it needn't be
renamed.

Does this seem reasonable?

[Jakob Kummerow, 2013-03-21 16:28:32]

Note on timing: we will not be able to land this before the Chromium M27 branch.
This CL *looks* safe, but it's too late for potentially-disruptive changes to
the build system before the branch point.

https://codereview.chromium.org/12790011/diff/1/tools/gyp/v8.gyp
File tools/gyp/v8.gyp (right):

https://codereview.chromium.org/12790011/diff/1/tools/gyp/v8.gyp#newcode46
tools/gyp/v8.gyp:46: 'dependencies': ['v8_base.<(v8_target_arch)',
'v8_snapshot'],
I would have assumed that v8_snapshot needs the <(v8_target_arch) treatment too.
Is that not the case?

https://codereview.chromium.org/12790011/diff/1/tools/gyp/v8.gyp#newcode52
tools/gyp/v8.gyp:52: 'dependencies': ['v8_base.<(v8_target_arch)',
'v8_nosnapshot.<(v8_target_arch)'],
nit: please respect 80-col lines where possible (breaking strings is not
necessary).

[Torne, 2013-03-21 17:00:25]

We are probably okay with this not landing before M27; I understand not wanting
to potentially break builds at this point. :)

https://codereview.chromium.org/12790011/diff/1/tools/gyp/v8.gyp
File tools/gyp/v8.gyp (right):

https://codereview.chromium.org/12790011/diff/1/tools/gyp/v8.gyp#newcode46
tools/gyp/v8.gyp:46: 'dependencies': ['v8_base.<(v8_target_arch)',
'v8_snapshot'],
On 2013/03/21 16:28:32, Jakob wrote:
> I would have assumed that v8_snapshot needs the <(v8_target_arch) treatment
too.
> Is that not the case?

v8_snapshot is not actually built for host in a normal build, so it's not really
necessary (mksnapshot doesn't depend on it). If we rename *everything* that can
be built for host then that would end up being the whole lot including libv8,
which you rightly wanted to avoid.

https://codereview.chromium.org/12790011/diff/1/tools/gyp/v8.gyp#newcode52
tools/gyp/v8.gyp:52: 'dependencies': ['v8_base.<(v8_target_arch)',
'v8_nosnapshot.<(v8_target_arch)'],
On 2013/03/21 16:28:32, Jakob wrote:
> nit: please respect 80-col lines where possible (breaking strings is not
> necessary).

Oops, yes. I shall reformat to 80 cols where reasonably possible.

[Torne, 2013-03-21 17:14:59]

Re-wrapping done. I will just leave this here until you're ready to land it :)

[Torne, 2013-04-02 11:29:04]

Are we sufficiently past the branch point to land this now?

[Jakob Kummerow, 2013-04-10 10:42:08]

On 2013/04/02 11:29:04, Torne wrote:
> Are we sufficiently past the branch point to land this now?

As of https://groups.google.com/d/msg/v8-dev/RxQEqu1o56w/EhkrIJtvYiYJ, we are.

LGTM. Landing.

[Jakob Kummerow, 2013-04-10 10:43:03]

Committed patchset #2 manually as r14209 (presubmit successful).