Rudy Lippan wrote:
> I'd like to find a way to test for this early on? Like checking defines
> for both libs or something. Thoughts?
I would think, that this is too difficult to detect, in particular among
architectures. (Solaris is not the only operating system around. :-)
My original idea, which I still find good, is that the "myld" script should
be able to parse the linker output for messages like the above and give
hints. IMO that's completely sufficient.
> I see that, but I don't see why, off hand. I am going to need to start
> digging.. Maybe ld is still returning 0 in this case? Wait... Looking at
> the message above again, I just noticed that it is a warning message, so I
> am going to assume that ld is not returining a non-zero status, and
> therefore, myld does not know that there is a problem.
Most probably you are right, and the "myld" script should change its
behaviour to forwarding all output immediately to stdout and only printing a
final diagnostic message, if that seems to make sense.