Skip to content
Commits on Jul 4, 2011
  1. Merge branch 'frontend'

    committed Jul 4, 2011
  2. checking in a file to test the constraintSolver. the file out.txt inc…

    …identally
    
    is also the same one against which the evaluation was run (eventually :p) for
    the 706 class project
    committed Jul 4, 2011
  3. adding some obfuscated c code. i believe this broke the c-interface. …

    …this would
    
    at least give a good starting point of stuff left in c. if it's mostly esoteric
    stuff, or that which isn't important then you can move to resurrecting some of
    the c++ stuff you'd gotten implemented (and moving on with that)
    committed Jul 4, 2011
  4. updating .gitignore

    committed Jul 4, 2011
  5. removing file which was used for debugging the frontend from the main

    repository. this should be included in the tests directory and/or probably
    broken down into sub-parts
    committed Jul 4, 2011
  6. Merge branch 'frontend'

    committed Jul 4, 2011
  7. Merge branch 'driver'

    committed Jul 4, 2011
  8. checking in the modified test that coonce had submitted. iirc he just…

    … added a
    
    bunch of obfuscated code to the original test. this is the test which was
    finally used to give some tangible numbers (previous ones were too small).
    
    perhaps a similar strategy can be used to generate more tests.
    committed Jul 4, 2011
  9. this is what the pr22061-3.c file should look like. this file is a pa…

    …rt of suite
    
    which were used to force crashes in earlier versions of gcc
    committed Jul 4, 2011
Commits on Jun 30, 2011
  1. Merge branch 'llvm-2.9'

    committed Jun 30, 2011
  2. Making the code runable under llvm-2.9.

    Previous merge with branch 'llvm-2.9' should have been a non-fastforward
    one. goofed up on that. Anyhoo, the code now at least compiles under the latest
    llvm release
    committed Jun 30, 2011
  3. Updating libs

    committed Jun 30, 2011
  4. constness checks

    committed Jun 30, 2011
  5. The Diagnostic class has been refactored since clang revision

    [[shell:svn%20log%20http://llvm.org/svn/llvm-project/cfe/trunk@119730][119730]]
    and a new abstraction of DiagnosticIDs was introduced. updating the method calls
    appropriately.
    committed Jun 30, 2011
  6. As per

    [[https://code.google.com/p/include-what-you-use/issues/detail?id%3D13][this]]
    it's no longer necessary to call setLLVMContext as a default LLVMContext is now
    constructed by CodeGenAction
    committed Jun 30, 2011
Commits on Jan 4, 2011
Commits on Jan 3, 2011
Commits on Dec 30, 2010
Commits on Dec 25, 2010
  1. Committing minor fix. I think I'm stopping now with this report. It's…

    … just a
    
    total piece of shit and needs to be rewritten. On the bright side i never knew
    plane rides could be so productive.
    committed Dec 24, 2010
  2. Removed obvious errors and fallacies. Fixed the Search strategy and e…

    …valuation
    
    sections.
    
    After having fixed the problems in the constraint generator which were keeping
    us from trying larger (and more complex) files i ran a comparison of our
    algorithm with ddmin (run on the same syntactic granularity). Not only did our
    algorithm converge in half the time and using an order of magnitude fewer tests,
    it's output was a third the size of the minimizing test case got by running
    ddmin. This was done using randomized search because nothing else was
    feasible. Even with randomized search the fact that even though we were able to
    reduce the number of tests by a factor of 10 and yet reduce the time only by a
    factor of 2 points to the need for a better inference system than the one we
    currently have. I might try a database solution before writing one myself. If i
    do write my own though i think imma use it to learn Haskell.
    committed Dec 24, 2010
  3. Added a weighted average search strategy. Also had forgotten to commi…

    …t a utility
    
    function that the random search depended on. Can't be arsed to go back and
    correct that now.
    committed Dec 24, 2010
Commits on Dec 24, 2010
  1. Was doing math wrong.

    committed Dec 24, 2010
Something went wrong with that request. Please try again.