Help make Base compile under NaCl...

Help make Base compile under NaCl

This patch works around some more missing APIs in NaCl.  This patch is kind of
gross.  A better approach might be to implement this missing function in NaCl
itself.

Original patch by Eric Seidel.

[eseidel, 2010-12-01 07:09:31]

On 2010/12/01 07:07:55, abarth wrote:

I *really* appreciate you upstreaming these!  I'm not sure this patch "works"
yet though.  I know we have some timer issues with the unit tests.  Until we
pass the timing unit tests I'm suspect of this patch.

[brettw, 2010-12-01 07:27:20]

On Tue, Nov 30, 2010 at 11:09 PM,  <eseidel@chromium.org> wrote:
> On 2010/12/01 07:07:55, abarth wrote:
>
> I *really* appreciate you upstreaming these!  I'm not sure this patch
> "works"
> yet though.  I know we have some timer issues with the unit tests.  Until we
> pass the timing unit tests I'm suspect of this patch.

I can't even understand the hack. If you guys agree we need to do it
this way, can it at least be commented well?

[macdome_gmail.com, 2010-12-01 07:29:19]

On 2010/12/01 07:27:20, brettw wrote:
> On Tue, Nov 30, 2010 at 11:09 PM,  <mailto:eseidel@chromium.org> wrote:
> > On 2010/12/01 07:07:55, abarth wrote:
> I can't even understand the hack. If you guys agree we need to do it
> this way, can it at least be commented well?

The hack is that nacl is missing timegm/gmtime.  There is a bug on that which we
should mention here.

The hack is just trying to emulate the behavior of those gnulib functions. 
again this will go away if NaCl gnulibc works.

[brettw, 2010-12-01 07:32:11]

On Tue, Nov 30, 2010 at 11:29 PM,  <macdome@gmail.com> wrote:
> On 2010/12/01 07:27:20, brettw wrote:
>>
>> On Tue, Nov 30, 2010 at 11:09 PM,  <mailto:eseidel@chromium.org> wrote:
>> > On 2010/12/01 07:07:55, abarth wrote:
>> I can't even understand the hack. If you guys agree we need to do it
>> this way, can it at least be commented well?
>
> The hack is that nacl is missing timegm/gmtime.  There is a bug on that
> which we
> should mention here.
>
> The hack is just trying to emulate the behavior of those gnulib functions.
> again this will go away if NaCl gnulibc works.

Yeah, I'm saying I think the setenv stuff should be clearer and
documented why you're doing this sequence of steps.

Brett

[macdome_gmail.com, 2010-12-01 07:52:15]

On 2010/12/01 07:32:11, brettw wrote:
> On Tue, Nov 30, 2010 at 11:29 PM,  <mailto:macdome@gmail.com> wrote:
> > On 2010/12/01 07:27:20, brettw wrote:
> >>
> >> On Tue, Nov 30, 2010 at 11:09 PM, &nbsp;<mailto:eseidel@chromium.org>
wrote:
> >> > On 2010/12/01 07:07:55, abarth wrote:
> >> I can't even understand the hack. If you guys agree we need to do it
> >> this way, can it at least be commented well?
> >
> > The hack is that nacl is missing timegm/gmtime. &nbsp;There is a bug on that
> > which we
> > should mention here.
> >
> > The hack is just trying to emulate the behavior of those gnulib functions.
> > again this will go away if NaCl gnulibc works.
> 
> Yeah, I'm saying I think the setenv stuff should be clearer and
> documented why you're doing this sequence of steps.
> 
> Brett

http://code.google.com/p/nativeclient/issues/detail?id=1160 is the NaCl bug
(which they closed as won't fix).  However that bug will be fixed automagically
if/when NaCl moves to glibc.

[brettw, 2010-12-01 22:59:56]

http://codereview.chromium.org/5434001/diff/1/base/time_posix.cc
File base/time_posix.cc (right):

http://codereview.chromium.org/5434001/diff/1/base/time_posix.cc#newcode66
base/time_posix.cc:66: //
http://code.google.com/p/nativeclient/issues/detail?id=1160
I'd like to at least see a comment here about why this works. Like along the
lines of "since we want GMT, we reset the time zone informations stored in the
environment, get the time, and then put it back. TODO(abarth) this will not work
across threads, blah blah blah."