Re: "unable to detect exception model"

[Paolo Carlini <pcarlini at unitus dot it>]

Hi!

thank you very much for the technical clarification!
(I had learned the hard way some of that...)

I think that in the meanwhile (waiting for the new eh in the branch) I will
try the "frankenstein-ish" move: I *must* understand if the cin problem I'm
experiencing with recent Gcc3.1 snapshots is newer-libstdc++-proper-related,
new eh-related, or what!

Is there something you suggest? Would you perhaps find useful a transcript of
a gdb session, single stepping into the library calls (starting a test code
with the input redirected from a file) ???

P.


Benjamin Kosnik wrote:

> > What I'm doing wrong?? It is really impossible to build a working
> > libstdc++-v3 without the new exception model?? In other terms: it is
> > presently impossible to test a new incarnation of libstdc++-v3 by using
> > "only" Gcc3.0 ??
>
> Short answer: yes.
>
> Long answer: not easily.
>
> The libsupc++ bits are pretty tied to gcc-3.1 (CVS head) until the
> call-frame exception handling bits get merged in.
>
> What you can try is using the old libsupc++ (in libstdc++-v3/libsupc++)
> on the gcc-3 branch with the rest of the libstdc++-v3 from gcc-3.1 or
> libstdc++-2.92.
>
> Sounds kind of frankenstein-ish, and it is. We need to merge in to gcc-3
> branch with this stuff.
>
> -benjamin