[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: gmp-4.0.1 and .so numbers

Dale Scheetz <dwarf@polaris.net> writes:

> On Sun, 3 Mar 2002, Josip Rodin wrote:

>> (It's rather doubtful that they really kept compatibility and bumped the
>> major version, but hey...)
> But hey... that was what bothered me about the whole thing. The
> whole point of naming major releases of the package differently was
> to provide two potentially incompatible versions of the library at
> the same time.  Now, what looks like a simple upgrade may very well
> break existing software by not providing a consistant ABI.

According to the NEWS file for gmp 4, it does state: 

* Upwardly binary compatible with 3.x versions.

There are lots of additions, newly supported platforms, new
algorithms, etc, which may account for the new major version number.
It looks to me like this is a case where they carefully managed
changes to keep ABI compatibility, but definitely added enough to
count.  In other words, as crazy as it sounds, it looks like they did
it right!

Maybe you could check with upstream to make sure.

> My typical test of such things is to install the new version and try ssh
> to my ISP. If that works then I scp the package to Incoming. This doesn't
> really feel like much of a test. Is it sufficient for this question?

If you want a more comprehensive test, here an idea (don't know if
it'll work, but it's worth a try).  Maybe you could compile up the
tests from the gmp 3.1 version, but sub in the libraries from 4.0.  If
make check succeeds, it's probably good.

Alan Shutko <ats@acm.org> - In a variety of flavors!
You should go home.

Reply to: