On Tue, 30 Oct 2001, Richard Levitte - VMS Whacker wrote: > Ah, good point... So you're saying that avoiding the symlinks > libcrypto.so.0 and libssl.so.0 kind of sends a clearer messgage, > right?
Yes, they would only be used for real, i.e. by the dynamic loader, if the respective SONAMEs were libcrypto.so.0 and libssl.so.0. Otherwise they are only used as an intermediate step in dereferencing the libcrypto.so and libssl.so symlinks used by the static linker. In this case, the symlinks may be set directly to their file targets; actually there are library installation recipes that do that in all cases -- there is a marginal run-time performance advantage here. I believe hardly anyone who compiles third party libraries, especially in bulk, pays much attention to how SONAMEs and symlinks relate to each other and it is well known that "lib*.so" symlinks are a part of development support but "lib*.so.*" symlinks are a part of run-time support. I've seen packages broken (conflicting between different versions) because of this assumption. Maciej -- + Maciej W. Rozycki, Technical University of Gdansk, Poland + +--------------------------------------------------------------+ + e-mail: [EMAIL PROTECTED], PGP key available + ______________________________________________________________________ OpenSSL Project http://www.openssl.org Development Mailing List [EMAIL PROTECTED] Automated List Manager [EMAIL PROTECTED]
