Ulrich Drepper writes:
> Richard Gooch <rgooch@xxxxxxxxxxxxxxx> writes:
> > The problem lies in the (lack of) documentation.
> Then why don't you write any if you want to see it?
??? The problem was in the lack of documentation when the code was
written! Thus, your claim of people not being able to programme was
And in my previous message I sent you a suggested line that could be
added to /usr/include/dlfcn.h to clarify things. Hopefully that will
> > Furthermore, under libc 5 there *is no* RTLD_DEFAULT, so it's hardly
> > surprising that RTLD_NEXT is what people have been using. Does glibc
> > still claim to be binary compatible with libc 5?
> What are you talking about? Glibc never was compatible with libc 5.
Really? I recall statements that libc 5-compiled programmes should be
able to work with glibc. Perhaps I'm mistaken. I do recall that glibc
2.0 was supposed to be compatible with glibc 2.1, but I note that
hasn't been the case (both source and binaries).
> > Finally, why don't you have error checking in your library? If the
> > usage is in fact wrong, isn't is cleaner to return NULL from dlsym(),
> > rather than having it segfault?
> You completely misunderstand the concept of a runtime library. It
> is assumed that programs are correct and have to be executed as fast
> as possible. There are no checks for programmer errors.
So when you changed the behaviour, you couldn't afford a simple if
test? Compared to all the other work that's being done, it's a trivial
amount of code.
You didn't actually break this intentionally, did you?