How widespread is this time-conversion bug?
Dennis Ritchie reported on comp.lang.c.moderated that Plan 9's gmtime() didn't believe that 2000 was a leap year. The problem is that it was working with tm_year instead of tm_year + 1900. Peter Seebach reported that BSD/OS had the same problem. I wonder when these bugs were added. The tz code looks fine. I thought BSD/OS was using tz; did they switch to another library? ---Dan
On 6 Feb 1999, D. J. Bernstein wrote:
Dennis Ritchie reported on comp.lang.c.moderated that Plan 9's gmtime() didn't believe that 2000 was a leap year. The problem is that it was working with tm_year instead of tm_year + 1900.
FWIW BSD pom(6) (phase-of-Moon) had the same problem until I reported it last July (now fixed in {Net,Open,Free}BSD). -- Joseph S. Myers jsm28@cam.ac.uk
Quoth D. J. Bernstein on Sat, Feb 06, 1999:
I wonder when these bugs were added. The tz code looks fine. I thought BSD/OS was using tz; did they switch to another library?
BSD/OS uses a very old version of tz. I integrated tz 1999b into BSD/OS 4.0, and will soon send BSDI the patches. Vadik. -- It was state of the art, he said. The art in this case was probably pottery. -- Terry Pratchett and Neil Gaiman, "Good Omens"
participants (3)
-
D. J. Bernstein -
Joseph S. Myers -
Vadim Vygonets