Re: [PATCH 3/6] Introduce a new "printf format" for timestamps

Johannes Schindelin [off-list ref] writes:

So let's introduce the pseudo format "PRItime" (currently simply being
"lu") so that it is easy later on to change the data type to time_t.

The problem being solved is a good thing to solve, and 

-	printf("author-time %lu\n", ci.author_time);
+	printf("author-time %"PRItime"\n", ci.author_time);

is one of the two ingredients to the solution for this line.  But
the final form would require casting ci.author_time to the type
expected by %PRItime format specifier.  With this change alone, you
could define PRItime to expect an winder type in the next step but
that would be a bad conversion.  IOW, changing only the format
without introducing an explicit cast appears to invite future
mistakes.

It would be better to introduce the timestamp_t we discussed earlier
before (or at) this step, and typedef it to ulong first, and then in
this step, change the above to

	printf("author-time %"PRItime"\n", (timestamp_t)ci.author_time);

to keep them in sync.  And at a later step in the series, you can
update definition of PRItime and timestamp_t to make them wider at
the same time, and the changes in this patch like the above line
would not need to be touched again.