Bug 14472
| Summary: | date for Y#K | ||
|---|---|---|---|
| Product: | Base System | Reporter: | Jin Guojun <jin> |
| Component: | bin | Assignee: | Sheldon Hearn <sheldonh> |
| Status: | Closed FIXED | ||
| Severity: | Affects Only Me | ||
| Priority: | Normal | ||
| Version: | 3.3-RELEASE | ||
| Hardware: | Any | ||
| OS: | Any | ||
On Fri, 22 Oct 1999 19:32:02 MST, Jin Guojun wrote:
> # date 199910221913.15
> date: illegal time format
The following patch should give you what you want. Note that you can
only specify 4-digit years within the range 1902-2037.
Ciao,
Sheldon.
Index: date.c
===================================================================
RCS file: /home/ncvs/src/bin/date/date.c,v
retrieving revision 1.30
diff -u -d -r1.30 date.c
--- date.c 1999/08/27 23:13:59 1.30
+++ date.c 1999/10/25 11:27:20
@@ -174,7 +174,9 @@
exit(retval);
}
-#define ATOI2(ar) ((ar)[0] - '0') * 10 + ((ar)[1] - '0'); (ar) += 2;
+#define ATOI2(ar) (((ar++)[0] - '0') * 10 + ((ar++)[1] - '0'))
+#define ATOI4(ar) (((ar++)[0] - '0') * 1000 + ((ar++)[1] - '0') * 100 + \
+ ((ar++)[2] - '0') * 10 + ((ar++)[3] - '0'))
void
setthetime(fmt, p, jflag, nflag)
const char *fmt;
@@ -221,11 +223,16 @@
/* if p has a ".ss" field then let's pretend it's not there */
switch (strlen(p) - ((dot != NULL) ? 3 : 0)) {
+ case 12: /* cc */
+ lt->tm_year = ATOI4(p) - 1900;
+ if (lt->tm_year < 0)
+ badformat();
+ goto year_done;
case 10: /* yy */
lt->tm_year = ATOI2(p);
if (lt->tm_year < 69) /* hack for 2000 ;-} */
lt->tm_year += 100;
- /* FALLTHROUGH */
+year_done: /* FALLTHROUGH */
case 8: /* mm */
lt->tm_mon = ATOI2(p);
if (lt->tm_mon > 12)
On Mon, 25 Oct 1999 13:28:07 +0200, Sheldon Hearn wrote:
> The following patch should give you what you want. Note that you can
> only specify 4-digit years within the range 1902-2037.
I sent you a diff for the source, but not for the manpage. The diff for
the latter follows.
Ciao,
Sheldon.
Index: date.1
===================================================================
RCS file: /home/ncvs/src/bin/date/date.1,v
retrieving revision 1.31
diff -u -d -r1.31 date.1
--- date.1 1999/08/27 23:13:58 1.31
+++ date.1 1999/10/25 11:49:14
@@ -49,7 +49,7 @@
.Op Fl t Ar minutes_west
.Op Fl v Ns Ar [+|-]val Ns Op ymwdHMS
.Ar ...
-.Op Fl f Ar fmt Ar date | [[[[yy]mm]dd]HH]MM[\&.ss]
+.Op Fl f Ar fmt Ar date | [[[[[cc]yy]mm]dd]HH]MM[\&.ss]
.Op Cm + Ns Ar format
.Sh DESCRIPTION
.Nm Date
@@ -74,7 +74,7 @@
.Ar fmt
as the format string to parse the date provided rather than using
the default
-.Ar [[[[yy]mm]dd]HH]MM[.ss]
+.Ar [[[[[cc]yy]mm]dd]HH]MM[.ss]
format. Parsing is done using
.Xr strptime 3 .
.It Fl j
@@ -168,6 +168,8 @@
The canonical representation for setting the date and time is:
.Pp
.Bl -tag -width Ds -compact -offset indent
+.It Ar cc
+Century (either 19 or 20) prepended to the abbreviated year.
.It Ar yy
Year in abbreviated form (e.g. 89 for 1989, 06 for 2006).
.It Ar mm
Thanks, the patch works. > > # date 199910221913.15 > > date: illegal time format > > The following patch should give you what you want. Note that you can > only specify 4-digit years within the range 1902-2037. I tried to tracing downto the time2 in /usr/src/lib/libc/stdtime/localtime.c but had no clue how it works. That is, I could not find how the 2037 is set. Any idea how we can fix this limitation? Thanks, -Jin > Ciao, > Sheldon. > > Index: date.c > =================================================================== On Mon, 25 Oct 1999 09:12:37 MST, jin@george.lbl.gov wrote: > That is, I could not find how the 2037 is set. Any idea how we > can fix this limitation? It's not a small job. :-) You'd need to change the definition of _BSD_TIME_T, which is a long at the moment. I imagine there's an enormous amount of software (including parts of the kernel) that'd choke on this. Ciao, Sheldon. PS: The function you'd wanna look at is settimeofday. BDE pointed out unsafe use of the post-increment operator in my diff.
This one should be safe.
Ciao,
Sheldon.
Index: date.c
===================================================================
RCS file: /home/ncvs/src/bin/date/date.c,v
retrieving revision 1.30
diff -u -d -r1.30 date.c
--- date.c 1999/08/27 23:13:59 1.30
+++ date.c 1999/10/27 07:41:09
@@ -175,6 +175,8 @@
}
#define ATOI2(ar) ((ar)[0] - '0') * 10 + ((ar)[1] - '0'); (ar) += 2;
+#define ATOI4(ar) ((ar)[0] - '0') * 1000 + ((ar)[1] - '0') * 100 + \
+ ((ar)[2] - '0') * 10 + ((ar)[3] - '0'); (ar) += 4;
void
setthetime(fmt, p, jflag, nflag)
const char *fmt;
@@ -221,11 +223,16 @@
/* if p has a ".ss" field then let's pretend it's not there */
switch (strlen(p) - ((dot != NULL) ? 3 : 0)) {
+ case 12: /* cc */
+ lt->tm_year = -1900 + ATOI4(p);
+ if (lt->tm_year < 0)
+ badformat();
+ goto year_done;
case 10: /* yy */
lt->tm_year = ATOI2(p);
if (lt->tm_year < 69) /* hack for 2000 ;-} */
lt->tm_year += 100;
- /* FALLTHROUGH */
+year_done: /* FALLTHROUGH */
case 8: /* mm */
lt->tm_mon = ATOI2(p);
if (lt->tm_mon > 12)
State Changed From-To: open->suspended Committed in date.c rev 1.31 and date.1 rev 1.32, waiting for MFC before closing. Responsible Changed From-To: freebsd-bugs->sheldonh TThis reminder is for me. :-) State Changed From-To: suspended->closed Merged from CURRENT in time for 3.4-RELEASE. |
`date` command does not take thousand/hundred digits. Since we are fixing Y2K problem, why don't we just simply fix YnK problem at once, so no one has to fix such problem in the next century, and for the rest centuries :-) How-To-Repeat: # date 199910221913.15 date: illegal time format usage: date [-nu] [-d dst] [-r seconds] [-t west] [-v[+|-]val[ymwdHMS]] ... [-f fmt date | [[[[yy]mm]dd]HH]MM[.ss]] [+format] # date 9910221913.30 Fri Oct 22 19:13:30 PDT 1999 # uname -rsm FreeBSD 3.3-RELEASE i386 # date 210001010100 date: illegal time format usage: date [-nu] [-d dst] [-r seconds] [-t west] [-v[+|-]val[ymwdHMS]] ... [-f fmt date | [[[[yy]mm]dd]HH]MM[.ss]] [+format] # date 0110221913.30 Mon Oct 22 19:13:30 PDT 2001 # date 9910221917.40 Fri Oct 22 19:17:40 PDT 1999