Re: libtool, java woes

[Jeff Sturm <jsturm at one-point dot com>]

On Mon, 9 Apr 2001, Bryce McKinlay wrote:
> > 3) Is it a good idea long-term to continue this interdependence between
> > libstdc++ and libjava?  It is a little unusual in that the other language
> > runtimes are independent.
> 
> What interdependence? Do you mean libsupc++? Because the java runtime is
> implemented (partly) in C++, it seems reasonable for it to depend on part of
> the C++ runtime. If it didn't, parts of the C++ runtime would have to be
> duplicated in libjava. This would create a dependency between the
> C++ compiler itself and libjava. I think that would be much worse! Generally,
> changes to libsupc++ shouldn't have any effect on libjava. The recent
> problems are just build related afaik.

Maybe so, but my thinking was that libsupc++ changes aren't likely to be
tested adequately if people aren't retesting java (and don't configure
with threads anyway).

Both the java and c++ frontends will depend on a frozen ABI so I'm
not sure the dependence between the frontends is an issue.

That's a small point though.  I'm frustrated that of the three platforms I
test with gcj, only one is building correctly at the moment.  And the
branch is what, almost two months old?  It's an indication of the
complexity of GCC these days that nobody is generally at fault
when something breaks... most patches I've seen were tested in good faith
by their contributors.  How large can GCC realistically grow while its
components are so tightly coupled?

Jeff