perm filename T2[AP,DBL] blob
sn#104890 filedate 1974-06-05 generic text, type T, neo UTF8
00100 1. Review
00200 2. Implementation details
00300 Precisely what knowledge you need to write induction programs
00400 How to actually construct a working system: Protocol
00500 Deviations from the theory: confessions
00600 Program parameters
00700 3. Questions one should ask about any automatic programming system
00800 Does it write one program?
00900 What is the input dialog like?
01000 Does it write a second program?
01100 How difficult is it to extend it to write more programs?
01200 Where inside it is the program really stored?
01300 4. Results
01400 Input: the user-pup dialogue
01500 Output: the CF program it produced
01600 5. A second task is chosen: grammatical inference
01700 The program it should write
01800 What new beings, functions, and changes to old PUP5 system are
01900 required? How hard will this new task be?
02000 6. Extensions
02100 What programs could be written with only minor modifications?
02200 What programs are very poorly suited to PUP5?
02300 7. Conclusions
02400 Comparison of PUP5 to other systems: size and time of system and of
02500 target programs. Relative performance.
02600 Disussion of effort expended and it worth, for each phase of the
02700 project: conception, protocol, programming, debugging,
02800 future target programs
02900 Was it the right way to solve the problem?
03000 Was it the right problem?
03100 Future directions
03200 immediate: more target programs, by more users
03300 theoretical: analysis of class of programs well/poorly
03400 suited to attack by PUP5 ideas.
03500 sometime: clean up system, more faithful to the ideas.
03600 destiny: what is the impact and the future of the ideas
03700 our group: coordination with other systems
03800 ease
03900 utility
04000 who's in control
04100 do you really want a ten page sort program?