Sorry, Harold. I've been really busy today, and I'm a DLL dunce. I've been trying to wrap my head around the problem so as to offer something useful.
I understand the basic issues involved and could see the problem after the lesstif list responded with that FAQ entry. But, I don't know enough about how/why DLLs require all symbols to be resolved to offer a work around. I'll keep looking... In the mean time, if we could produce a list of the symbols involved, that would make things go smoother when we do find a solution. On Mon, 27 Oct 2003, Harold L Hunt II wrote: > Brian, > > Sounds like we are going to have more trouble with Xt and LessTif. > Torrey Lyons sent me a great description of the problem on OS X (I > believe I forwarded it to the list). It boils down to Xt having some > unresolved symobls that we resolve at link time by linking to ICE and SM > (these libraries provide the default functions). However, other > libraries are supposed to be able to override these functions with their > own implementations, which the linker is expected to figure out. As we > all know, DLLs cannot have unresolved symbols, thus this functionality > is lost. We would need to provide an alternative to this behavior or > attempt to modify the linker to provide this functionality. > > In the meantime this means that the shared build of Xt doesn't work. > This is probably causing some of the weird problems people have noticed > with keyboard focus. This also explains why some types of LessTif apps > don't work anymore. > > We should probably revert to a static Xt and LessTif for the time being. > I know, this is not desireable, but it is probably going to take > months for this to be fixed and we can't have broken libs and apps for > the duration. > > Thoughts? > -- Brian Ford Senior Realtime Software Engineer VITAL - Visual Simulation Systems FlightSafety International Phone: 314-551-8460 Fax: 314-551-8444
