We review selected technical challenges addressed in our digital 
library project. Our InfoBus, a CORBA-based distributed object 
infrastructure, unifies access to heterogeneous document collections and
information processing services. We organize search access using a 
protocol (DLIOP) that is tailored for use with distributed objects. 
A metadata architecture supports novel user interfaces and query 
translation facilities. We briefly explain these components and then 
describe how technology choices such as distributed objects, commercial 
cataloguing schemes and Java, helped and hindered our progress. We also 
describe the evolution of our design tradeoffs.