Re: [ANNOUNCE] Git v2.9.1

[Johannes Schindelin <Johannes.Schindelin@gmx.de>]

Hi Peff,

On Tue, 12 Jul 2016, Jeff King wrote:

> On Tue, Jul 12, 2016 at 01:25:20PM +0200, Johannes Schindelin wrote:
> 
> > [PATCH] Work around test failures due to timestamps being unsigned long
> > 
> > Git's source code refers to timestamps as unsigned longs. On 32-bit
> > platforms, as well as on Windows, unsigned long is not large enough to
> > capture dates that are "absurdly far in the future".
> > 
> > While we will fix this issue properly by replacing unsigned long ->
> > time_t, on the maint track we want to be a bit more conservative and
> > just skip those tests.
> > 
> > Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
> > ---
> 
> This looks like a reasonable interim fix for both Windows and for the
> general maint track. If we reliably produce "2038" for the 999... value
> then that's simpler than adding in magic to ask about sizeof(ulong). I
> also wondered if we should test forthe _correct_ value, but that would
> defeat the point of the test (999... is also far in the future).

This was just a quick fix, intended to allow me to release Git for Windows
v2.9.1 in a timely manner (which is now delayed for other reasons).

I think I can do even better than to inspect the source code whether it
reliably clamps the dates to PI hours on January 19th, 2038: We already
have a t/helper/test-date and we can easily teach it one new trick.

> One minor comment:
> 
> > diff --git a/t/t0006-date.sh b/t/t0006-date.sh
> > index 04ce535..d185640 100755
> > --- a/t/t0006-date.sh
> > +++ b/t/t0006-date.sh
> > @@ -48,10 +48,17 @@ check_show default "$TIME" 'Wed Jun 15 16:13:20 2016 +0200'
> >  check_show raw "$TIME" '1466000000 +0200'
> >  check_show iso-local "$TIME" '2016-06-15 14:13:20 +0000'
> >  
> > -# arbitrary time absurdly far in the future
> > -FUTURE="5758122296 -0400"
> > -check_show iso       "$FUTURE" "2152-06-19 18:24:56 -0400"
> > -check_show iso-local "$FUTURE" "2152-06-19 22:24:56 +0000"
> > +case "$(test-date show:iso 9999999999)" in
> > +*2038*)
> > +	# on this platform, unsigned long is 32-bit, i.e. not large enough
> > +	;;
> > +*)
> > +	# arbitrary time absurdly far in the future
> > +	FUTURE="5758122296 -0400"
> > +	check_show iso       "$FUTURE" "2152-06-19 18:24:56 -0400"
> > +	check_show iso-local "$FUTURE" "2152-06-19 22:24:56 +0000"
> > +	;;
> > +esac
> 
> It would be nice to wrap this in a prereq, like:
> 
>   test_lazy_prereq 64BIT_TIME '
> 	case "$(test-date show:short 9999999999)" in
> 	*2038*)
> 		false
> 		;;
> 	*)
> 		true
> 		;;
> 	esac
>   '
> 
>   ...
>   check_show 64BIT_TIME iso       "$FUTURE" "2152-06-19 18:24:56 -0400"
>   check_show 64BIT_TIME iso-local "$FUTURE" "2152-06-19 22:24:56 +0000"
> 
> The main advantage is that it will number the tests consistently, and
> give us a "skipped" message (which can remind us to drop the prereq
> later when everything goes 64-bit).
> 
> But it also will do the right thing with test-date's output with respect
> to "-v" (not that we expect it to produce any output).
> 
> You'll need to adjust check_show as I did in my earlier patch.

Makes sense!

Thanks,
Dscho