msvs: fix regression in r1378, environment not set up properly for multiple actions on one input

msvs: fix regression in r1378, environment not set up properly for multiple actions on one input

The previous method (only setup on first iteration of loop) didn't respect
that the actions might be distributed across multiple input files, which are
run as separate commands. Instead, only do setup_env for first action for a
given input.

R=grt@chromium.org,jeanluc@chromium.org
BUG=gyp:261
Committed: https://code.google.com/p/gyp/source/detail?r=1381

[scottmg, 2012-05-16 21:56:46]

Try this again...

[jeanluc1, 2012-05-16 22:11:06]

You have tested it for both VS2008 and VS2010?

https://chromiumcodereview.appspot.com/10387164/diff/4001/pylib/gyp/generator...
File pylib/gyp/generator/msvs.py (right):

https://chromiumcodereview.appspot.com/10387164/diff/4001/pylib/gyp/generator...
pylib/gyp/generator/msvs.py:1460: have_setup_env = {}
Please document what this variable does (e.g. key/value)

[scottmg, 2012-05-16 22:20:46]

On 2012/05/16 22:11:06, jeanluc1 wrote:
> You have tested it for both VS2008 and VS2010?

I have now. :)

>
https://chromiumcodereview.appspot.com/10387164/diff/4001/pylib/gyp/generator...
> File pylib/gyp/generator/msvs.py (right):
> 
>
https://chromiumcodereview.appspot.com/10387164/diff/4001/pylib/gyp/generator...
> pylib/gyp/generator/msvs.py:1460: have_setup_env = {}
> Please document what this variable does (e.g. key/value)

Done. (Made it a set() based on your comment, since there's no useful value)

[jeanluc1, 2012-05-16 22:25:36]

lgtm

[jeanluc1, 2012-05-16 22:26:49]

On 2012/05/16 22:25:36, jeanluc1 wrote:

Could only some of the actions be triggered?  Thus missing the initial one if it
does not need to be run?

[scottmg, 2012-05-16 22:29:49]

On 2012/05/16 22:26:49, jeanluc1 wrote:
> On 2012/05/16 22:25:36, jeanluc1 wrote:
> 
> Could only some of the actions be triggered?  Thus missing the initial one if
it
> does not need to be run?

I don't think so. The whole set gets attached as one command line (the original
problem) so it should always be at the beginning this way. (As far as I
understand it anyway.)

[jeanluc1, 2012-05-16 22:49:18]

On 2012/05/16 22:29:49, scottmg wrote:
> On 2012/05/16 22:26:49, jeanluc1 wrote:
> > On 2012/05/16 22:25:36, jeanluc1 wrote:
> > 
> > Could only some of the actions be triggered?  Thus missing the initial one
if
> it
> > does not need to be run?
> 
> I don't think so. The whole set gets attached as one command line (the
original
> problem) so it should always be at the beginning this way. (As far as I
> understand it anyway.)

In your test, after the first run, you could probably touch some of the source
files and see if it behaves correctly.

[scottmg, 2012-05-16 23:07:12]

On 2012/05/16 22:49:18, jeanluc1 wrote:
> On 2012/05/16 22:29:49, scottmg wrote:
> > On 2012/05/16 22:26:49, jeanluc1 wrote:
> > > On 2012/05/16 22:25:36, jeanluc1 wrote:
> > > 
> > > Could only some of the actions be triggered?  Thus missing the initial one
> if
> > it
> > > does not need to be run?
> > 
> > I don't think so. The whole set gets attached as one command line (the
> original
> > problem) so it should always be at the beginning this way. (As far as I
> > understand it anyway.)
> 
> In your test, after the first run, you could probably touch some of the source
> files and see if it behaves correctly.

https://chromiumcodereview.appspot.com/10383222/

[jeanluc1, 2012-05-16 23:13:47]

LGTM

On 2012/05/16 23:07:12, scottmg wrote:
> On 2012/05/16 22:49:18, jeanluc1 wrote:
> > On 2012/05/16 22:29:49, scottmg wrote:
> > > On 2012/05/16 22:26:49, jeanluc1 wrote:
> > > > On 2012/05/16 22:25:36, jeanluc1 wrote:
> > > > 
> > > > Could only some of the actions be triggered?  Thus missing the initial
one
> > if
> > > it
> > > > does not need to be run?
> > > 
> > > I don't think so. The whole set gets attached as one command line (the
> > original
> > > problem) so it should always be at the beginning this way. (As far as I
> > > understand it anyway.)
> > 
> > In your test, after the first run, you could probably touch some of the
source
> > files and see if it behaves correctly.
> 
> https://chromiumcodereview.appspot.com/10383222/