1. CSC 313 – Advanced Programming Topics

2. Today’s Goal  Make you forget reading that was assigned  I went back & reviewed others; none like this one  Highlight process used by modern compilers  Consider steps taken & how design works  Where optimizations occur will be discussed  Review a kick-ass design honed through years  Understand decisions to see good design in action  See design process that results in design patterns

3. Translations  Translating docs from English to other language  But all of this material selected from college library  Undecipherable & meaningless documents included  Workers highly-trained & so saving time (& $$) critical  Reject untranslatable documents & explain why  Need to give workers chance to correct mistakes  Money important, so limit time on these documents

4. Translations  Translating docs from English to other language  But all of this material selected from college library  Undecipherable & meaningless documents included  Workers highly-trained & so saving time (& $$) critical  Reject untranslatable documents & explain why  Need to give workers chance to correct mistakes  Money important, so limit time on these documents How can we do this?

5. What Errors Can Occur?

6.  Document read in unreadable or not English  Has “words”, but grammatically incorrect  Meaningless jumble means cannot translate

7. What Errors Can Occur?  Document read in unreadable or not English αυηθι μςτ ϕϖ κλσροπ χψτν γδζεεθ μςτυ  Has “words”, but grammatically incorrect  Meaningless jumble means cannot translate

8. What Errors Can Occur?  Document read in unreadable or not English αυηθι μςτ ϕϖ κλσροπ χψτν γδζεεθ μςτυ  Has “words”, but grammatically incorrect 4 u 2 c I sent U pix. ty  Meaningless jumble means cannot translate

9. What Errors Can Occur?  Document read in unreadable or not English αυηθι μςτ ϕϖ κλσροπ χψτν γδζεεθ μςτυ  Has “words”, but grammatically incorrect 4 u 2 c I sent U pix. ty  Meaningless jumble means cannot translate Mary flow over the swimming pizza

10. How To Translate?

12. Money, money, money!  This project useful for many languages  Different alphabet or punctuation may be used  Will need to adjust to different grammar rules  Change in scoping rules & variable declaration types

13. Money, money, money!

14.  This project useful for many languages  Different alphabet or punctuation may be used  Will need to adjust to different grammar rules  Change in scoping rules & variable declaration types  Could translate to many different languages  (Unless written by Microsoft or in Alabama)

15. How To Translate?

16. How To Translate? (Redux)  Split translations process into multiple layers  Lexical analyzer will check legal words & punctuation  Grammar errors found by parser (syntactic analysis)  Semantic analysis will check document meaningful  Optimizations happen here, since program is valid  Code generator “lowers” results into new language

17. All About the Benjamins  Most layers’ actions language-dependent  Reusing details hard, but algorithms very similar  Between layers, pass language-independent data  Define protocols specifying how layers interact  Each layer independent & can change easily  Limit rewriting, instead reuse whenever possible  Must follow protocols, but these set out clearly  Popular approach: gcc, XCode, Google Translate

18. Planning Data & Interactions  Data to be passed depends on work layers do  Individual words & punctuation output by lexer  Parser goes through this data & find sentences  Build paragraphs by checking sentences meaningful  Interactions based on input size needed  Item-by-item analyses needed to build sentences  Semantic analysis works by using whole module  Entire program needed for optimizations & output

19. Planning Data & Interactions  Data to be passed depends on work layers do  Individual words & punctuation output by lexer  Parser goes through this data & find sentences  Build paragraphs by checking sentences meaningful  Interactions based on input size needed iteration  Item iteration needed to build sentences  Semantic analysis works by using whole module  Entire program needed for optimizations & output

20. Resulting Design  Lexer is an Iterator for words & punctuation  Illegal characters found & reported as errors  Parser builds syntax trees for each module  Function, class, or file may be language’s module  Must be conservative: outputs result for entire file  Good syntax tree output by semantic analysis  When use found, variable linked to declaration & type  Match calls with functions & all checked if legal HIR  Result called Higher-order Intermediate Representation

21. Lexical Analysis In Detail  Creates “tokens” from file being compiled  While it does this, checks characters are legal  Keywords identified ( for, int, while …) in input  Code parsed to identify literals (strings, numbers)  Does not understand code; only splitting it out  Only layer looking at code, so skips comments

22. How Does Lexer Work?  Regular expressions defined for items to find  They’re back!! Turned into big finite state machine  Defines enum or hierarchy for returned data struct Symbol { Token type; String value; } enum Token { NUMBER, STRING, WORD, GT,,... }

23. Why Use State Machines?  Limits knowledge needed for coding a lexer  Machine has set of states connected by edges  State determines next state, given an input  Need to know initial & accepting states only  States define language; processing unchanged

24. Why Use State Machines?  Limits knowledge needed for coding a lexer  Machine has set of states connected by edges  State determines next state, given an input  Need to know initial & accepting states only  States define language; processing unchanged  Obviously, this is part of State design pattern (Covered later in term, in case you wondering)  Maximizes reuse of code within lexical analysis  Limits rework to developing regular expressions

25. For Next Lecture  Read pages 37 – 55 in book  Get back into easier code & design patterns  1-to-many communication is problem why?  Why is this critical to all event-based coding?  Lindsay Lohan's secret to success related to this?