TODO.txt

MAINTAIN

    Keep __all__ up to date with all the useful functions

    Comment and properly docstring everything

    Keep the -h follow all the other common -h practices

    Make sure that the -h is up to date, including all the version info of all the external features (gcc/Cython)

    Make sure that you constantly search for the assumptions cyther makes

    Make my getIncludeAndRuntime() function is up to date with build_ext in distutils

    Any name cool badges to add? Make sure you got all the recent and useful ones
    Use http://shields.io/ to generate non changing badges

    Make sure the test_cyther.py file is updated to include the latest features (test everything!)


TO-DO (before next version)

    The --test option
        Add this to the help text
        Say that if there is an error, to email me the output of this function
    What is the use of the vcruntime.dll on windows installations? Does it need to be included?
    Download many different ubuntu and linux or mac operating systems on virtual box to directly test
    Travis keeps telling me this: 'python: can't open file 'cytherize': [Errno 2] No such file or directory'
    Make sure a new window doesnt get spawned on the subprocess calls (on some systems it does, some it doesnt)
    Implement the makefile system
        But for now make sure it created a file with just commands
    Make sure that 'cytherize' can be called from the command line correctly
    Where is the best spot to put the README and all the other files?
    Fix the help text in the readme to not have 'Nothing here yet'
        Is there a better way to get the version of gcc?
    Change the descripting of the readme so that it doesn't say that it is directly a compiler, but a processor instead

EASY FIX

    Whats up with CYTHER_CONFIG_FILE? Should we use it?

MEDIUM FIX

    Fix issue posted on cyther/issues
        Investigate cold installations and usage of other operating systems

HARD FIX

    Get the operating system dependant flags that are passed to gcc by distutils and inject them when building the
    preset (see question you posed)

EASY NEW FEATURES

    Get the arguments of run(timer=True) to work properly, or be automatic, like IPython

    Get crawl to find libpython, and also rename crawl to 'what'


HARD NEW FEATURES

    Erase Cyther's usage of distutils entirely

    Instead of the include option, have the system automatically detect the python {}.get_include() modules

    Say the search for the directories necessary for the runtime compilation fails, and returns nothing. Then run the
    crawl algorithm over the whole DRIVE to find a lib file that matches the REGEX definition


LOW PRIORITY

    Function to clean a directory of everything 'locally build' related, or move it to a cythercache
        A function to remove EVERYTHING cyther related

    Always back up the previous compile if there was one. Then if there is an error, don't overwrite it with anything,
    but instead, keep the backup version. Have the new compilation attempt be built in the __cythercache__ directory

########################################################################################################################

BEFORE DEPLOY (for me)

    Make sure that the version specified makes sense (major.minor.patch)

    Make sure that the README reflects the new interface changes

    Make sure that the CHANGELOG is up to date and includes the current version change

    Delete the comments that are no longer needed. Any junk or temporary commented code should be extracted

    Write a test script that actually tests the whole deal. Have it run every scenario and then
    USE the actual extension that it makes


QUESTIONS (for me to ask myself)

    How to deal with non-standard extensions that people might be using? Don't error, but just have it print that
    it is non standard, and work anyway? What about Regex?

    Don't have args and file, have only one object with everything in it. Does this even make sense?

    Is it wise to take out the checking procedures with the getIncludeAndRuntime() function. Why are they necessary??

    Would it be possible to automatically send me their __cytherinfo__ variable?

    Will sift ever have to return multiple results for the lib directories??

    What about lines like this platform.platform().strip().lower().startswith('windows')??
    What do the platform names even look like? Is their structure reliable? Does it need to be?

    Is it necessary to make sure that os.pardir will never be a problem??

    What about packaging the libraries with cyther?? Is cross compiling possible / a good idea?

    What is the difference between os.name and sys.platform and platform.platform?