Am 28.02.2017 um 15:28 schrieb Jeff King:

On Mon, Feb 27, 2017 at 10:30:20PM +0100, Johannes Schindelin wrote:

quoted

One notable fallout of this patch series is that on 64-bit Linux (and
other platforms where `unsigned long` is 64-bit), we now limit the range
of dates to LONG_MAX (i.e. the *signed* maximum value). This needs to be
done as `time_t` can be signed (and indeed is at least on my Ubuntu
setup).

Obviously, I think that we can live with that, and I hope that all
interested parties agree.

I do not just agree, but I think the move to a signed timestamp is a big
improvement. Git's object format is happy to represent times before
1970, but the code is not. I know this has been a pain for people who
import ancient histories into Git.

It looks from the discussion like the sanest path forward is our own
signed-64bit timestamp_t. That's unfortunate compared to using the
standard time_t, but hopefully it would reduce the number of knobs (like
TIME_T_IS_INT64) in the long run.

Glibc will get a way to enable 64-bit time_t on 32-bit platforms 
eventually (https://sourceware.org/glibc/wiki/Y2038ProofnessDesign). 
Can platforms that won't provide a 64-bit time_t by 2038 be actually 
used at that point?  How would we get time information on them?  How 
would a custom timestamp_t help us?

Regarding the need for knobs: We could let the compiler chose between 
strtoll() and strtol() based on the size of time_t, in an inline 
function.  The maximum value can be calculated using its size as well. 
And we could use PRIdMAX and cast to intmax_t for printing.

René