Re: New control field proposal which could help on gcc3.2 transition
On Thu, 12 Sep 2002, Ben Collins wrote:
> I program can more easily retain backward compatibility though. You've
> generally done so quite well by having things like an API where the user
> of the API defines which version it is using.
> Libraries can do the same thing with versioned symbols (and I've yet to
> understand why more things don't do this...portability perhaps).
So far, what I gathered from the library symbol versioning stuff is:
1. GNU systems have absolutely no reason not to use it.
2. We should be using and enforcing its usage.
3. We don't because the other distros don't, and they don't because of the
Oh, and since the soname is used as the versioning (plus c++ internal ABI
and whatever, if we're lucky), soname screwups will end up harder to clean
I also have this feeling that should we start using symbol versioning
widely, we could get gcc upstream to help us add the interal ABI to it...
"One disk to rule them all, One disk to find them. One disk to bring
them all and in the darkness grind them. In the Land of Redmond
where the shadows lie." -- The Silicon Valley Tarot