Add a support for newlib

Add a support for newlib

Newlib has _timezone instead of __timezone. pNaCl with newlib needs this change.

BUG=132145
TEST=When rolled, NaCl (with newlib) can be compiled without an error.
R=mark@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=267335

[jungshik at Google, 2014-04-30 10:24:58]

Can you take a look? Thanks


I misinterpreted pnacl.h.patch for ICU 46 and didn't port it to ICU 52. That led
to a buildbot failure when I tried to roll ICU to ICU 52. 

This should fix it. An upstream bug about newlib was filed as noted in
README.chromium. 

BTW, I wonder if there's a way to "positive-id' newlib (instead of !__GLIBC__).

[Mark Mentovai, 2014-04-30 14:06:27]

LGTM. Does this work?

https://codereview.chromium.org/266503003/diff/1/README.chromium
File README.chromium (right):

https://codereview.chromium.org/266503003/diff/1/README.chromium#newcode29
README.chromium:29: - patches/putil.patch for Android,  QNX and
newlib(NaCl-newlib).
extra space before QNX

https://codereview.chromium.org/266503003/diff/1/source/common/putilimp.h
File source/common/putilimp.h (right):

https://codereview.chromium.org/266503003/diff/1/source/common/putilimp.h#new...
source/common/putilimp.h:119: /* uClibc does not have __timezone nor _timezone.
*/
or, not nor.

https://codereview.chromium.org/266503003/diff/1/source/common/putilimp.h#new...
source/common/putilimp.h:124: #      else
https://codereview.chromium.org/162913002/ uses _NEWLIB_VERSION. You can try
that since you were looking for a way to test for “is newlib” instead of ”isn’t
glibc.”

I might get rid of a nesting level here, as in:

#   if defined(__UCLIBC__)
       /* uClibc does not have __timezone or _timezone. */
#   elif defined(_NEWLIB_VERSION)
#      define U_TIMEZONE _timezone
#   else
#      define U_TIMEZONE __timezone
#   endif

[jungshik at Google, 2014-04-30 21:18:53]

Committed patchset #2 manually as r267335 (presubmit successful).

[jungshik at Google, 2014-04-30 22:05:48]

On 2014/04/30 14:06:27, Mark Mentovai wrote:
> LGTM. Does this work?
> 
> https://codereview.chromium.org/266503003/diff/1/README.chromium
> File README.chromium (right):
> 
> https://codereview.chromium.org/266503003/diff/1/README.chromium#newcode29
> README.chromium:29: - patches/putil.patch for Android,  QNX and
> newlib(NaCl-newlib).
> extra space before QNX
> 
> https://codereview.chromium.org/266503003/diff/1/source/common/putilimp.h
> File source/common/putilimp.h (right):
> 
>
https://codereview.chromium.org/266503003/diff/1/source/common/putilimp.h#new...
> source/common/putilimp.h:119: /* uClibc does not have __timezone nor
_timezone.
> */
> or, not nor.
> 
>
https://codereview.chromium.org/266503003/diff/1/source/common/putilimp.h#new...
> source/common/putilimp.h:124: #      else
> https://codereview.chromium.org/162913002/ uses _NEWLIB_VERSION. You can try
> that since you were looking for a way to test for “is newlib” instead of
”isn’t
> glibc.”
> 
> I might get rid of a nesting level here, as in:
> 
> #   if defined(__UCLIBC__)
>        /* uClibc does not have __timezone or _timezone. */
> #   elif defined(_NEWLIB_VERSION)
> #      define U_TIMEZONE _timezone
> #   else
> #      define U_TIMEZONE __timezone
> #   endif

Thank you for the review/suggestion and figuring out a newlib-id method for me.

Landed with the suggested change.

[jungshik at Google, 2014-04-30 22:22:24]

For the record,  the test to do is 'ninja -C out/Release base_i18n_nacl'. 

There's one more issue and it's taken care of at
https://codereview.chromium.org/263763002/