Re: nostalgia: libc5 -> libc6 transition
martin f krafft <firstname.lastname@example.org> writes:
> I apologise for bringing up things of the past, but I am researching
> and need your help. Back when I upgraded Bo to Hamm, it worked like
> a charm. Ever since then, I have heard the "Debian was the only one
> to handle the libc6 migration gracefully" many, many times.
> Now, after researching quite a bit, it seems that the transition was
> not so smooth after all. For instance, apparently X broke (which
> I didn't notice as I wasn't using it), and third-party non-Debian
> software didn't run anymore (due to -rpath pointing at the location
> where the libc6 files had replaced the libc5 ones). I imagine that
> if Hamm also provided updated X libraries, then the first problem
> was easily fixed whereas the second could be worked around with
> a hacked ld.so.
> So is it true to say that Debian handled the transition nicely?
Debian handled the transition nicely. Broken third party software
didn't. There is nothing Debian could have done to make third party
> Also, along the same lines, I'd be interested to hear what about
> Debian and the policy made the transition possible. Right now,
> I have -rpath and SONAME-change-on-ABI-change noted, but both are
> really lessons we learned from the libc6 transition, not something
> that enabled it, right?
Shlibs files that set the correct Depends on updated libraries.
> Looking forward to your input!
Maybe the gcc-3.2 transition would be a better case since that
included what was learned from the libc6 transition.