less is not reproducible when built from git

[bbhtt]

It's generating timestamps that cause the issue.

See

1. less/mkhelp.pl

   Lines 10 to 12 in e77e117

   	my ($sec,$min,$hour,$mday,$mon,$year) = gmtime();
   	printf "/* This file was generated by mkhelp.pl from less.hlp at %d:%02d on %d/%d/%d */\n",
   	$hour, $min, $year+1900, $mon+1, $mday;

2. less/mkhelp.py

   Lines 6 to 9 in e77e117

   	time = time.gmtime()
   	print("/* This file was generated by mkhelp.py from less.hlp at "\
   	"%d:%02d GMT on %d/%d/%d */\n" %
   	(time.tm_hour, time.tm_min, time.tm_year, time.tm_mon, time.tm_mday))

3. less/Makefile.aut

   Lines 92 to 98 in e77e117

   	REPLACE_VERSION = \
   	DT=`date '+%d %h %Y'`; \
   	echo "Stuffing version number ${REL} into $@"; \
   	rm -f $@; \
   	sed \
   	-e "s;@@VERSION@@;${REL};" \
   	-e "s;@@DATE@@;$$DT;" >$@

Diffoscope:

I temporarily patched it to remove timestamp generation https://gitlab.com/freedesktop-sdk/freedesktop-sdk/-/merge_requests/22374, I don't think manpages need them.

[gwsw]

I think this is the same issue as #567. That was fixed in c02f755. But that fix doesn't remove timestamps from the man pages. It seems to me that most man pages do have timestamps. On my Fedora system I looked at 1277 man pages, and 987of them have a timestamp in the last line. It doesn't seem like a good idea for less's man page to be different from most of the others in this respect. But perhaps that timestamp could be changed to be baed on SOURCE_DATE_EPOCH as well.

[bbhtt]

The fix is not on master so I didn't notice it but yea using that for man pages too is probably also fine.

[avih]

On my Fedora system I looked at 1277 man pages, and 987of them have a timestamp in the last line. It doesn't seem like a good idea for less's man page to be different from most of the others in this respect.

Agreed, but someone which needs a reproducible build have other priorities.

But perhaps that timestamp could be changed to be baed on SOURCE_DATE_EPOCH as well.

Yeah, sounds good to me too.

[bbhtt]

Any idea if these fixes will go in master or later in release tags? I can't tell if the pos659 branch has a special significance.

[gwsw]

post659 was a development branch while the 668 bug fix release was being developed in master. Now that 668 is released, post659 will be merged back into master, probably in a week or so.

[gwsw]

Fixed in 98d5fd4.