Re: [HACKERS] "24" < INT_MIN returns TRUE ??? - Mailing list pgsql-hackers

From Tom Lane
Subject Re: [HACKERS] "24" < INT_MIN returns TRUE ???
Date
Msg-id 2462.931532532@sss.pgh.pa.us
Whole thread Raw
In response to Re: [HACKERS] "24" < INT_MIN returns TRUE ???  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: [HACKERS] "24" < INT_MIN returns TRUE ???
List pgsql-hackers
I said:
> Do you have <limits.h>, and if so how does it define INT_MIN?

Actually, looking closer, it doesn't matter whether you have <limits.h>,
because there is yet a *third* bug in numutils.c:
#ifdef HAVE_LIMITS#include <limits.h>#endif

should be
#ifdef HAVE_LIMITS_H...

because that is how configure and config.h spell the configuration
symbol.  Thus, <limits.h> is never included on *any* platform,
and our broken default INT_MIN is always used.

Whoever wrote this code was not having a good day...
        regards, tom lane


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: [HACKERS] "24" < INT_MIN returns TRUE ???
Next
From: Louis Bertrand
Date:
Subject: Re: [HACKERS] Hashing passwords (was Updated TODO list)