On Thu, 23 Sep 1999, Christoph Badura wrote:

> fb@enteract.com (Frederick Bruckman) writes:
> >> On Wed, 22 Sep 1999, Frederick Bruckman wrote:
> >> > - - Disable build on NetBSD-1.3[A-J] (needs libstdc++).
> >Moreover, it works on 1.4.1, without libg++, and it's reported to
> >build on 1.3.X.
> 
> >I only noticed this after kim reported a failure on 1.3H.
> 
> Hmm.  What exactly makes you believe that that is a NetBSD problem and not
> Kimmo's local problem? :-)

Intuition? The error message showed that the linker failed to find
very many symbols. What else could it be?

> >I'm not positive exactly when libstdc++ appeared. (ISTR 1.3K)
> 
> Uhm, 1.3.2 shipped with libstdc++.so.0.1 (and associated files).
> And zhadum (1.3I, supposedly) has:
> 
> % ls -l /usr/lib/libstdc*
> -r--r--r--  1 root  wheel  322044 Jan 20  1999 /usr/lib/libstdc++.a
> -r--r--r--  1 root  wheel  301927 Aug 31  1998 /usr/lib/libstdc++.so.0.1
> -r--r--r--  1 root  wheel  260531 Oct 28  1998 /usr/lib/libstdc++.so.1.0
> -r--r--r--  1 root  wheel  306043 Jan 20  1999 /usr/lib/libstdc++.so.2.0
> -r--r--r--  1 root  wheel  351932 Jan 20  1999 /usr/lib/libstdc++_p.a
> -r--r--r--  1 root  wheel  356630 Jan 20  1999 /usr/lib/libstdc++_pic.a
> 
> What version is good enough, I don't know.

I found an old 1.3H volume, and it has only libstdc++.so.1.0, Oct 23,
1998. Also a 1.3I, libstdc++.so.2.0, Feb 4, 1999, same as you.  So
probably 1.3I is OK. That's also consistent with the idea the 1.3H
doesn't have the right lib. It's downright puzzling that the package
is reported to build on 1.3.2 (but does it work?)

Now that I'm looking at it again, that wildcard is screwy, since it
permits 1.3, 1.3.1, 1.3.2, and 1.3.3. OTH, maybe it really is
libstc++.so.1.0 that's broken, and 0.1 is OK for this. I'd prefer to
collect reports of success/failure before trying to fix that.

If anyone with 1.3[A-J] wants to try this and report, it's trivial to
"fake out" the package system with "make OS_VERSION=1.4" or similar.