Re: binaries compiled on slink don't run anywhere else
----- Original Message -----
From: Raul Miller <rdm@test.legislate.com>
To: <debian-devel@lists.debian.org>
Sent: Saturday, November 14, 1998 8:20 PM
Subject: Re: binaries compiled on slink don't run anywhere else
>Dale Scheetz <dwarf@polaris.net> wrote:
>> Binary packages compiled on Hamm don't run on Bo, nor do all Bo packages
>> run on Rex. When you build packages, you often have to make sure that the
>> libraries you build with are the same as the ones on the target machine.
>>
>> 2.0.7u declares its lack of backward compatibility with shlibdeps, and by
>> conflicting with early sysklogd versions.
>
>Er.. if I recall correctly, rex was a.out (libc4), bo was H.J.Lu's elf
>(libc5) and hamm was glibc (libc6). Also, we did retain the ability
>to compile binaries for a previous version of the libraries. Hamm did
>include a libc5 development system, for example.
Nope, buzz (1.1) was elf/libc5, the only a.out releases of debian were the
prereleases 0.9x's and not even all of those were a.out, as the elf
conversion was underway (much like the libc5->libc6 conversion). The reason
for no 1.0 release, was because InfoMagic accidently published a CD with
Debian 1.0 on it, though that was a developer's copy and didn't even install
correctly, so instead of furthering the confusion, Debian released 1.1 as
their first release. After that came 1.2 (rex) and 1.3 (rex), which were
also elf/libc5. After that came 2.0 (hamm) being the hard switch to
glibc2/libc6.
On the issue of being able to compile for other's distributions, I believe
someone has already mentioned that's it's not neccisarily a glibc issue, but
a compiler issue, which might be able to be fixed with an additional -l flag
passed to the compiler, has anyone investigated this?
Shaya
Reply to: