Ensure GYP_PARALLEL prints error messages.

Ensure GYP_PARALLEL prints error messages.

Errors sent to stdout were being swallowed by gclient runhooks; moved them
to stderr. Also added an informational message when there's an error, and
made it so that GYP_PARALLEL=0 disables parallel processing.

After this change, we should be able to turn parallel on by default.

BUG=154205

[thakis, 2012-12-21 19:45:53]

https://codereview.appspot.com/7000046/diff/1/pylib/gyp/input.py
File pylib/gyp/input.py (right):

https://codereview.appspot.com/7000046/diff/1/pylib/gyp/input.py#newcode2564
pylib/gyp/input.py:2564: 'To disable, set GYP_PARALLEL=0 in your environment.')
Is this necessary? I'd not print anything in the regular case.

[dmazzoni, 2012-12-21 19:51:55]

https://codereview.appspot.com/7000046/diff/1/pylib/gyp/input.py
File pylib/gyp/input.py (right):

https://codereview.appspot.com/7000046/diff/1/pylib/gyp/input.py#newcode2564
pylib/gyp/input.py:2564: 'To disable, set GYP_PARALLEL=0 in your environment.')
On 2012/12/21 19:45:53, thakis wrote:
> Is this necessary? I'd not print anything in the regular case.

The last thing I want to do is kill productivity if this breaks for somebody.
How about we flip the flag, then remove this after developers have been using it
for a couple of weeks without issues?

[thakis, 2012-12-21 19:53:55]

https://codereview.appspot.com/7000046/diff/1/pylib/gyp/input.py
File pylib/gyp/input.py (right):

https://codereview.appspot.com/7000046/diff/1/pylib/gyp/input.py#newcode2564
pylib/gyp/input.py:2564: 'To disable, set GYP_PARALLEL=0 in your environment.')
On 2012/12/21 19:51:55, dmazzoni wrote:
> On 2012/12/21 19:45:53, thakis wrote:
> > Is this necessary? I'd not print anything in the regular case.
> 
> The last thing I want to do is kill productivity if this breaks for somebody.
> How about we flip the flag, then remove this after developers have been using
it
> for a couple of weeks without issues?

If something breaks, the error output above will contain the same note, no?

[dmazzoni, 2012-12-21 19:55:46]

On 2012/12/21 19:53:55, thakis wrote:
> If something breaks, the error output above will contain the same note, no?

I was more worried about what happens if it just hangs, like it was doing on
Windows before we fixed it.

[thakis, 2012-12-21 20:05:08]

lgtm

On 2012/12/21 19:55:46, dmazzoni wrote:
> On 2012/12/21 19:53:55, thakis wrote:
> > If something breaks, the error output above will contain the same note, no?
> 
> I was more worried about what happens if it just hangs, like it was doing on
> Windows before we fixed it.

I'd remove it anyways. Without this note, you'll hear about issues like this.
With it, people will just randomly flip this flag when they see any issues,
which means you might not hear about problems, or there might be unrelated
problems that you hear about, etc.

It's fine when you remove this in a few weeks, but that's also what you said
about the existence of this flag :-/

[scottmg, 2012-12-22 04:04:37]

https://codereview.appspot.com/7000046/diff/5001/pylib/gyp/input.py
File pylib/gyp/input.py (right):

https://codereview.appspot.com/7000046/diff/5001/pylib/gyp/input.py#newcode576
pylib/gyp/input.py:576: 'If the error only occurs when GYP_PARELLEL=1, '
if you don't remove it, PARELLEL->PARALLEL