Re: Automated report: NetBSD-current/i386 test failure

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Sun, 28 Apr 2019 21:50:45 +0000 (UTC)
    From:        NetBSD Test Fixture <bracket%NetBSD.org@localhost>
    Message-ID:  <155648824563.27186.9065778287415893510%babylon5.NetBSD.org@localhost>

  | The newly failing test case is:
  |
  |     lib/libutil/t_snprintb:snprintb

I know what is happening with that, and I will fix...

The problem relates to a difference of opinion/misunderstanding
of the 'F' (new style) format operator

The man page says...

	 Describes a multi-bit field like `f', but just extracts the value

and when we look at 'f', what it says is ...

	Describes a multi-bit field [....]
	The remaining characters are printed ...

If F is just like f then it needs "remaining characters" even though
the F format doesn't use them.

That's the cause of the apparent "off by one" which seemed to be
causing the first : or = to be ignored ... it was being ignored
because it was being treated as those unused "remaining characters".

Since I doubt the objective right now is to alter how F format has
worked for a long time (eg: in the tests) I will restore the "off
by one" after which it all magically works again.

I will also change the new doc to indicate explicitly that these
remaining characters must be given (just the terminating \0 works).

Then I'll add some ATF tests for the new stuff.

kre