Migrating to Blueprint

First consider autonomous case, then consider distributed case

1. Message types are replaced by typed stores
2. Threads are mostly replaced by methods
3. GUI interfaces are replaced by Call-backs
4. 'Classes' are typically mapped to circuits and are therefore encapsulated at the same granularity
5. private functions are performed by sub-circuits
6. Irregular functions are parallelized through dependency analysis
7. Data-parallel calculations are best implemented using multi-dimensional objects
8. Re-usable components will typically subscribe to their consumed events (inputs) but own the public stores that provide their outputs.
9. Top-down versus bottom up.  Go for coarse grain first
10. Move file scope and static data

 

 

---

© Connective Logic Systems Ltd. All Rights Reserved.