TODOs

[ex0dus-0x]

(migrated from https://github.com/pkmoore/rrapper/pull/6)

Documentation:

• [x] Documentation for rrinit and rrtest in README.
• [x] Document cpuid.c code
• [x] Docstrings for inject.py and rreplay

Logging:

• [ ] Add logging mechanisms for rrtest.
• [ ] Unify logging mechanisms for inject and parse_syscall_definitions to use a basicConfig logger.

Features:

• [x] Support for rreplay or rrtest to list existing tests
• [ ] Mechanism for rrtest to pack tests for distribution and portability
• [x] Mechanism for rrinit to copy over rr tests to CrashSim configuration file
• [ ] More accurate trace line / ReplaySession comparison in rrtest
• [ ] Enhance setup.py such that requirements.txt do not need to exist
• [ ] Better cleanup routine for rreplay
• [x] Change trace_file to point to snippet file
• [x] Move all applicaton-generated files (i.e pickles, JSON, pipes, etc) into ~/.crashsim for internalization and portability (TODO: make posix-omni-parser compliant as well)
• [ ] Verbose and "clean" outputs + error messages (hmmmm.... unified CrashSimulator exception class in consts.py???)
• [x] Fix runtime IndexError / EOF when choosing events near end of file

Testing:

• [ ] Just testing in general, seeing whether there are any run-time errors.

[liyun-li]

$.02. Please let me know whether my is too naive.

The current version of rrapper generates a ton of potential events for larger projects. I personally find it pretty inconvenient having to go through 1,000 events. Maybe for a temporary solution, we can pass an option to rrtest to reduce the range by brute forcing the session with rreplay and weed out the ones that yield errors?

So maybe rrtest configure -n app -t 1234 --guess-event?

EDIT: seeing the new change in branch output_event_number. Nevermind!