Subject: Re: 1.12.maint1, second beta
To: dj AT ctron DOT com (DJ Delorie)
Date: Fri, 26 Aug 1994 09:35:56 +0200 (MET DST)
From: Henrik Storner <storner AT olicom DOT dk>
Cc: djgpp AT sun DOT soe DOT clarkson DOT edu

DJ writes:
> 
> > And while on this point, does anyone know where can I found the definition
> > of how stat() should use TZ?  Some POSIX paper, maybe?  Any site which
> 
> POSIX.1 states that for the time-related fields, "Times are given in
> seconds since the Epoch."  Epoch is defined as "00:00:00 Jan 1 1970
> Coordinated Universal Time."
> 
> I interpret CUT to be the same as GMT+0, so TZ would (or should) have
> no affect, except that DOS stores times as local time instead of CUT
> time.

Meaning ... what, exactly ? That the current DJGPP-stat() behaviour is
wrong, I suppose ? Surely, the values presently returned by stat() do
not fit in anywhere with the POSIX definition you mention; the value that
stat() gives is GMT-time+4 hours, presumably du to some hardwired setting
if EST5 as the timezone. Note that the values does _not_ change with the
TZ setting!

A simple fix would be to use the current stat() value, subtract 4 hours
and then convert from localtime to GMT. Cannot be all that hard, eh ?
Like this:

*** libsrc/c/sys/stat.org       Tue Jul 19 03:35:22 1994
--- libsrc/c/sys/stat.c Fri Aug 26 15:33:20 1994
***************
*** 80,83 ****
--- 80,84 ----
  {
      static int        stat_called_before = 0;
+     static int  stat_gmtfix = 0;
      char      p[900]; /* Should be p[PATH_MAX+1] */
      int               status;
***************
*** 89,94 ****
--- 90,104 ----
      if (!stat_called_before)
      {
+         /* Initialize the first time called */
+         time_t ltm, gtm;
+
        stat_called_before = 1;
        (void) time((time_t *) 0);
+
+         ltm = gtm = time (NULL);
+         ltm = mktime (localtime (&ltm));
+         gtm = mktime (gmtime (&gtm));
+       /* # of seconds GMT is offset from local */
+         stat_gmtfix = (int)(gtm - ltm) - 14400;
      }

***************
*** 138,141 ****
--- 148,156 ----
           stat_assist doesn't think so. */
        if (S_ISDIR(buf->st_mode)) buf->st_mode |= S_IWRITE;
+
+       /* Fixup times returned by stat() from being EST5-based */
+       buf->st_ctime += stat_gmtfix;
+       buf->st_mtime += stat_gmtfix;
+       buf->st_atime += stat_gmtfix;
      }
-- 
Henrik Storner        | "Man is the best computer we can put aboard a space-
(storner AT olicom DOT dk)   |  craft ... and the only one that can be mass produced 
System Engineering    |  with unskilled labor."
Olicom Denmark        |                                   Wernher von Braun