Matthias Schmitt <freak002@mmp.lu> writes:
> On 28.10.2013, at 17:22, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> What I find curious is that I can't
>> reproduce this problem on an OS X Mavericks machine here. You must
>> be using some nondefault compiler switches --- care to tell us what?
> ./configure CC='/usr/bin/gcc-4.2' --prefix=/<my path>/applications/pgsql --with-perl
Hm. There is no such file as /usr/bin/gcc-4.2 on my machine, in fact
there's no file following such a naming pattern anywhere on the
filesystem.
However, that doesn't seem to be relevant. Poking at this some more,
I see that the effect of the bit that Andres noticed in string.h is to
#define strncpy as __builtin___strncpy_chk, and *that replacement is
happening on my machine*. So why don't I see the trap? I am wondering
if there is some run-time environment test that controls whether the
trap occurs, and you've got an environment variable set that I don't.
regards, tom lane