Add gyp target to build preparser as stand-alone library.

Add gyp target to build preparser as stand-alone library.

Likely only works on Linux yet.

[Søren Thygesen Gjesse, 2010-12-09 14:08:00]

LGTM,

but I think you should try it in chromium to trin away the options which are
inferred. If you run build/gyp_chromium you should be able to run 'make
v8_preparser'.

http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp
File tools/gyp/v8.gyp (right):

http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode33
tools/gyp/v8.gyp:33: 'host_arch%': '<(target_arch)',
I don't think we should set host_arch here it should have been supplied by the
consumer of v8.gyp. When used in Chromium this should all be set up.

http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode200
tools/gyp/v8.gyp:200: 'target_name': 'v8-preparser',
- -> _, other targets use underscore.

http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode227
tools/gyp/v8.gyp:227: ['OS=="linux" and library=="shared_library"', {
I also think -fPIC should be provided by the consumer of v8.gyp. We don't set
-fPIC for any other targets in this file.

http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode230
tools/gyp/v8.gyp:230: ['v8_target_arch!="x64" and host_arch=="x64"', {
Does the preparser actually care about the v8_target_arch?

Also I think that the only case where v8_target_arch != target_arch makes sense
is when v8_target_arch is arm and target_arch is ia32 (independent of
host_arch).

[Søren Thygesen Gjesse, 2010-12-10 09:10:42]

On 2010/12/09 14:08:00, Søren Gjesse wrote:
> LGTM,
> 
> but I think you should try it in chromium to trin away the options which are
> inferred. If you run build/gyp_chromium you should be able to run 'make
> v8_preparser'.
> 
> http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp
> File tools/gyp/v8.gyp (right):
> 
> http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode33
> tools/gyp/v8.gyp:33: 'host_arch%': '<(target_arch)',
> I don't think we should set host_arch here it should have been supplied by the
> consumer of v8.gyp. When used in Chromium this should all be set up.
> 
> http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode200
> tools/gyp/v8.gyp:200: 'target_name': 'v8-preparser',
> - -> _, other targets use underscore.
> 
> http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode227
> tools/gyp/v8.gyp:227: ['OS=="linux" and library=="shared_library"', {
> I also think -fPIC should be provided by the consumer of v8.gyp. We don't set
> -fPIC for any other targets in this file.
> 
> http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode230
> tools/gyp/v8.gyp:230: ['v8_target_arch!="x64" and host_arch=="x64"', {
> Does the preparser actually care about the v8_target_arch?
> 
> Also I think that the only case where v8_target_arch != target_arch makes
sense
> is when v8_target_arch is arm and target_arch is ia32 (independent of
> host_arch).

On 2010/12/09 14:08:00, Søren Gjesse wrote:
> LGTM,
> 
> but I think you should try it in chromium to trin away the options which are
> inferred. If you run build/gyp_chromium you should be able to run 'make
> v8_preparser'.
> 
> http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp
> File tools/gyp/v8.gyp (right):
> 
> http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode33
> tools/gyp/v8.gyp:33: 'host_arch%': '<(target_arch)',
> I don't think we should set host_arch here it should have been supplied by the
> consumer of v8.gyp. When used in Chromium this should all be set up.
> 
> http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode200
> tools/gyp/v8.gyp:200: 'target_name': 'v8-preparser',
> - -> _, other targets use underscore.
> 
> http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode227
> tools/gyp/v8.gyp:227: ['OS=="linux" and library=="shared_library"', {
> I also think -fPIC should be provided by the consumer of v8.gyp. We don't set
> -fPIC for any other targets in this file.
> 
> http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode230
> tools/gyp/v8.gyp:230: ['v8_target_arch!="x64" and host_arch=="x64"', {
> Does the preparser actually care about the v8_target_arch?
> 
> Also I think that the only case where v8_target_arch != target_arch makes
sense
> is when v8_target_arch is arm and target_arch is ia32 (independent of
> host_arch).

I have just tested this in Chromium with the suggested changes (removing setting
host_arch, cflags and ldflags) and with GYP_DEFINES set to target_arch=ia32, and
it works like a charm

$ make v8_preparser
  CXX(target) out/Debug/obj.target/v8_preparser/v8/src/allocation.o
  CXX(target) out/Debug/obj.target/v8_preparser/v8/src/hashmap.o
  CXX(target) out/Debug/obj.target/v8_preparser/v8/src/preparse-data.o
  CXX(target) out/Debug/obj.target/v8_preparser/v8/src/preparser.o
  CXX(target) out/Debug/obj.target/v8_preparser/v8/src/preparser-api.o
  CXX(target) out/Debug/obj.target/v8_preparser/v8/src/scanner-base.o
  CXX(target) out/Debug/obj.target/v8_preparser/v8/src/token.o
  CXX(target) out/Debug/obj.target/v8_preparser/v8/src/unicode.o
  AR(target) out/Debug/obj.target/v8/tools/gyp/libv8_preparser.a

We should probably try without target_arch=ia32 on a x86 machine.

[Lasse Reichstein, 2010-12-13 09:19:49]

http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp
File tools/gyp/v8.gyp (right):

http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode33
tools/gyp/v8.gyp:33: 'host_arch%': '<(target_arch)',
Removed.

http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode200
tools/gyp/v8.gyp:200: 'target_name': 'v8-preparser',
On 2010/12/09 14:08:00, Søren Gjesse wrote:
> - -> _, other targets use underscore.

Done.

http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode227
tools/gyp/v8.gyp:227: ['OS=="linux" and library=="shared_library"', {
On 2010/12/09 14:08:00, Søren Gjesse wrote:
> I also think -fPIC should be provided by the consumer of v8.gyp. We don't set
> -fPIC for any other targets in this file.

Done.

http://codereview.chromium.org/5716001/diff/1/tools/gyp/v8.gyp#newcode230
tools/gyp/v8.gyp:230: ['v8_target_arch!="x64" and host_arch=="x64"', {
Removed.