bug#58224: 29.0.50; "make bootstrap" spuriously warns: "comp.el newer than byte-compiled file"

[Alan Mackenzie]

Hello, Eli.

On Sun, Oct 02, 2022 at 14:04:08 +0300, Eli Zaretskii wrote:
> > Date: Sun, 2 Oct 2022 10:43:44 +0000
> > Cc: stefankangas@gmail.com, 58224@debbugs.gnu.org, acm@muc.de
> > From: Alan Mackenzie <acm@muc.de>

> > > Instead of inventing a new value that overrides the non-nil value, why
> > > not simply reset the variable to nil?

> > force-load-messages is nil by default, and currently isn't used at all
> > by Emacs.  It seems to be a pure debugging variable.

> > The NOMESSAGE argument to Fload when non-nil, causes the unwanted
> > message:

> >     Source file `foo.el' newer than byte-compiled file; using older file

> > ..  When NOMESSAGE is nil, we get instead

> >     Loading foo.elc (compiled; note, source file is newer)...

> > ..  Whichever setting of NOMESSAGE and force-load-messages we use, we get
> > one of the above messages displayed.  So, I'm proposing using a new
> > value 'never for force-load-messages to mean display neither of these
> > messages.

> I don't want to complicate the public Lisp API because we have a
> singular situation at some point of the bootstrap, and for minor
> aesthetic reasons at that; that is the tail wagging the dog.  So let's
> fix this more subtly.

> How about recognizing (inside Fload) a specific time stamp of the
> older file we use (we set it to the beginning of the Epoch, right?),
> and suppressing the message in that case?

I've got this working, but ....

In lread.c I've got:

  struct timespec epoch_timespec = {(time_t)0, 0}; /* 1970-01-01T00:00 UTC */ 

, which clearly isn't satisfactory.  Can you (or anybody else) give me a
clue as how to convert a human readable time into a struct timespec?
I've spent most of the afternoon searching and grepping lots of .h files,
and haven't come up with anything, yet.

Thanks!

-- 
Alan Mackenzie (Nuremberg, Germany).