Roland Mainz wrote: > Rod Evans wrote: >> Mike Kupfer wrote: >>> Hey Rod, the ksh93 integration will introduce a few new libraries. >>> What's the timeframe for the "direct bindings" changes that you're >>> working on, and what does the ksh93 team need to know? >> I wouldn't worry about "direct bindings" too much, it's always >> been a moving target. I've a few pending integrations if I can >> even get out from under everyday fire drills. But I expect to >> be refining things within the link-editors and the OSNet contents >> as I work towards enabling direct binding. > > Erm, I hope this is not another tree-wide change which required to > update our tree...
No, there's nothing like that ... there are minor Makefile tweaks. > BTW: What are "direct bindings" ? They are a model of symbol lookup. A caller (say someone who references printf()) records the dependency of the supplier (libc.so.1) of the definition at link-edit time. Then at runtime, ld.so.1 directly binds the caller to the definition, ie. ld.so.1 does not undergo the default symbol search looking for printf(), which starts at the a.out and progresses through each loaded dependency until it finds a definition (which is typically libc). Your putback looks imminent. Mine's a ways off. -- Rod.
