On Mon, 9 Apr 2001, Van Duser, Brad wrote:
> > I just tried building 5.6.1 on R2.10 with -Dusedl and I found that the
> > build died when building the B extension. I found that I had
> > to issue:
> >
> > export _C89_CCMODE=1
> >
> > The failures I saw in the op/*.t tests included:
> >
> > op/bop...............FAILED at test 22
>
> Peter, just to clarify, do you mean that you saw these failures due to not
> having set CCMODE or you saw these failures even after building with
> _C89_CCMODE=1?
Those were the failures I saw even after building with _C89_CCMODE=1.
Having specified that I wanted Dynamic Loading with my test build via
-Dusedl I found that in order to get the build past the compilation of the
B extension that I had to have _C89_CCMODE=1 in my environment. I
conducted no test of DynaLoading perl without _C89_CCMODE=1.
I seemed to also have massive failures in the t/lib/*.t tests, but I am
refraining from saying too much about it until I can re-verify that I
wasn't up against some resource contention issue (entirely possible on
that particular LPAR).
If I recall correctly (today has been a very busy day on many fronts) I
did not see massive failures in t/lib/*.t tests with a static, that is a
default, build of Perl 5.6.1 on OS/390 2.10.
Peter Prymmer
