Home Original page

PATCH: Update libiberty/testsuite/demangle-expected

H. J. Lu hjl@lucon.org
Fri Oct 17 19:47:00 GMT 2003 
On Fri, Oct 17, 2003 at 02:41:15PM -0400, DJ Delorie wrote:
> 
> > I agree we shouldn't change those testcases which are passing. The
> > failed ones are just reminders for us how they should be fixed.
> 
> But even for failing cases, changing the expected output changes the
> functional specifications for the code.  That's never an obvious
> change.

There was never a functional specification for those failed tests. The
one in libiberty is copied from libstdc++.

> 
> > I don't agree syncing with the original working source is blindly
> > updating unless the original one is wrong.
> 
> If libiberty and libstdc++ shared sources for the demangler, and there
> was an agreement in place to keep them in sync, and the testsuite
> change was done in conjunction with a source change, and it was all
> discussed and approved for libstdc++, you'd be right.
> 
> But it's not.

Regardless if libiberty and libstdc++ share the demangler source or
not, should they have the same output on the same input for those
failed ones in libiberty? If not, why?


H.J.
More information about the Binutils mailing list