Subject: Re: mysql NOW() bug on Alpha?
To: Dave Huang <email@example.com>
From: Simon Burge <firstname.lastname@example.org>
Date: 10/07/2000 00:19:34
Dave Huang wrote:
> On Fri, 6 Oct 2000, Dave Huang wrote:
> > Hmm, interesting... I bet it has to do with negative timezone offsets...
> > if I link /etc/localtime to GMT or Australia, it works, but US doesn't.
> Yeah, that does seem to be the problem... when loading in the timezone
> info, it's stuffing the GMT offset into a 64-bit long rather than 32-bit
> int, so negative offsets are being treated as huge positive offsets. The
> patch below seems to fix the problem for me... perhaps the mysql-server
> package should be updated to use a more recent pthreads package rather
> than the one that comes with mysql? (devel/pth perhaps? or
> devel/unproven-pthreads? I dunno what the difference is between 'em :)
Good catch Dave!
One comment, shouldn't the "result" var in the detzcode() function
also be a "int32_t"?
> --- mit-pthreads/gen/ctime.c.orig Sun Oct 12 07:07:41 1997
> +++ mit-pthreads/gen/ctime.c Fri Oct 6 08:07:31 2000
> @@ -129,7 +129,7 @@
> ** Prototypes for static functions.
> -static long detzcode __P_((const char *));
> +static int32_t detzcode __P_((const char *));
> static const char * getnum __P_((const char *, int *, int, int));
> static const char * getsecs __P_((const char *, long *));
> static const char * getoffset __P_((const char *, long *));
> @@ -175,7 +175,7 @@
> time_t altzone = 0;
> #endif /* defined ALTZONE */
> -static long detzcode(const char * codep)
> +static int32_t detzcode(const char * codep)
> long result;
> int i;
Simon Burge <email@example.com>
NetBSD Sales, Support and Service: http://www.wasabisystems.com/