Hi Magnus, On Tue, May 15, 2007 at 11:05:40AM +0200, Magnus Holmgren wrote: > According to the nettle docs, summer needs be linked with -lnettle -lgmp, > but I don't know if this is Right. As it happens, I was just discussing > this with upstream.
No it's not right. No software should be required to care about the implementation details of the library it's using, *especially* when dynamic linking on an OS that gets this right (such as GNU/Linux). The previous upstream version of libnettle (or at least, the Debian packaging thereof) got this right. The current behavior is a regression. While I will concede that it's *possible* to make every dependent package adapt when the library changes the internal details of which libraries it depends on, it is nevertheless *wrong* to do so. It's completely non-scalable on a variety of levels, and in this particular case breaks various useful technologies such as prelink which rely on being able to sort out the dependencies of individual ELF libraries. This regression also breaks any packages (e.g., those from etch) built against the previous version of libnettle, because upgrading libnettle2 will leave these gmpz symbols unresolved -- the old versions of the package didn't need to link against libgmp, the new version of libnettle didn't, so nothing links to libgmp. Here again, it's obvious that the part that should be linking against libgmp is the part that's actually *using* it -- libnettle, not the random app. Thanks, -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. [EMAIL PROTECTED] http://www.debian.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
