wdibi / pivot Goto Github PK

someFunc(_ y) -> _
    if x then return z;
end

If x is always true return is reachable. If we don't know the value of x the return statement might not be reachable.

Particularly with lists and dicts

• Whenever we update our grammar we need to check that our README examples are still valid.
• Is there anyway to automate this?

// FAIL

firstFactorial(num x) -> num
    if x == 0 or x == 1 then
        return 1;
    else
        return x * firstFactorial(x - 1);
    end
end

// Pass

firstFactorial(num x) -> num
    if x == 0 or x == 1 then
        return 1;
    end
        return x * firstFactorial(x - 1);
end

• Grammar for indicating a matrix type.
• Matrix operator.
• Valid matrix semantic checks

• Project clones and runs without crashing
• All tests pass
• Your compiler has commandline operations that can run the parser, the analyzer, and the generator separately
• Can analyze and generate code for Hello World
• Can analyze and generate code for a medium size program
• Can detect lots of errors correctly
• Your language has some “difficult” (or at least highly non-trivial) features to analyze
• Code is nice, clean, and professional
• Unit tests cover all interesting language features that should pass
• Unit Tests for semantics - at least a dozen, preferably 30 or more, erroneous cases are covered
• Code generator is complete for your language
• Your language has a feature for which code generation was interesting and non-trivial
• Code generator code is relatively efficient (no glaring problems)
• Unit Tests for code generator - at least one
• Unit Tests for code generator - at least several
• Unit Tests for code generator - decent coverage
• README page and companion website are complete, useful, and accurate
• README page OR companion website has a link to the grammar (making it easy to lookup)
• Project commit log shows that you've been working all semester

• Implement directed edges
• Add edge labels
• Consider refactoring view(program) in pivot.js to return a Promise. Might not matter since nodemon is is running and will monitor changes.
• To avoid git, can we pass the graph data around another way? Or is our best option to add ./graph/data.json to the .gitignore?

• Project Can be Cloned
• All Tests Pass
• Can be run on commandline without -o and produces unoptimized code
• Can be run on commandline with -o and produces optimized code
• Repo contains several example programs of different complexities, that can be compiled
• I can run the linter from the commandline using npm run lint and there are NO errors
• 100% Test Coverage
• New Strength Reduction Optimizations
• TWO Non-trivial optimization that was not in Tiger
• You have documented your optimizations

Total Points: 87 (137 Points Total, 50 from Lightning Talk)

Instead of

temp <- a;
a <- a + b;
b <- temp;
x <- x - 1;

we should allow

temp, a, b, x <- a, a+b, temp, x-1;

equivalent is the wrong term for this and can be confusing.