11 Dec
2001
11 Dec
'01
09:03
On Tuesday 11 December 2001 09.41, Philipp Thomas wrote:
You should never symlink .1 to .2 unless you can make sure that all structures are identical, functions take the same arguments and so on. Being backwards compatible does not suffice, the binary interface must be identical.
Not identical, surely. But the .1 functions and structures must exist in .2, I realize that. I was more concerned that the system somehow tried to outsmart me in a win-esque sort of way, but I think I've understood it now. Thanks for the explanation in the other mail, BTW. //Anders