| From | Sent On | Attachments |
|---|---|---|
| Jakob Breivik Grimstveit | Mar 28, 2001 4:05 am | |
| Bob Schader | Mar 28, 2001 4:55 am | |
| Sam Varshavchik | Mar 28, 2001 5:25 am | |
| Sam Varshavchik | Mar 28, 2001 5:27 am | |
| Bob Schader | Mar 28, 2001 5:43 am | |
| Jakob Breivik Grimstveit | Mar 28, 2001 5:48 am | |
| Bob Schader | Mar 28, 2001 5:59 am | |
| Sam Varshavchik | Mar 28, 2001 11:46 am | |
| Sam Varshavchik | Mar 28, 2001 11:48 am | |
| Bob Schader | Mar 28, 2001 12:12 pm |
| Subject: | [courier-users] Re: Errors while checking | |
|---|---|---|
| From: | Sam Varshavchik (mrs...@courier-mta.com) | |
| Date: | Mar 28, 2001 11:46:51 am | |
| List: | net.sourceforge.lists.courier-users | |
Bob Schader writes:
My compiler is gcc 2.95.2 from sunfreeware, so I can't really put the blame on Sun. I had the same problem compiling MySQL, ehich also uses c++, that is where I found the answer to this problem in the first place.
The problem technically occurs in the runtime linker, which cannot find shared libraries at runtime. Sun's runtime linker is rather simplistic. It looks in /lib and /usr/lib, and that's it, pretty much hardcoding the paths. A much better approach is to cache the location if each library, which allows libraries to be installed in any location.
Although some people may claim this as a bug, I personally see no reason why the proper testing and configuring could not be added to the configure script. Thats what its there for, right?
Perhaps, but this is such a basic module that it is reasonable to expect the OS to handle it. We are not talking about some custom non-standard, manually, installed. We're talking about the C++ runtime library, which is really a part of the operating system, more than anything else. With such a basic system component you shouldnt' really have to jump through hoops in order to use it. It should work without having to use any special procedure to load it.
-- Sam





