- Finally fixed an age-old bug in unit tests. The bug was that whenever a test
output contained '*', shell globbing expanded that to a list of all test
files. Even though it was simply diagnostic output text.
- Set well-known umask for permissions test. When started with another
umask value the permissions of the created directory would be
different and the tests fail.
- Removed the \n characters from Task::composeFF4 and Task::ComposeJSON.
- Tasks in backlog.data file are now written as JSON.
- Tasks are recognized in FF4 or JSON format.
- Many files were missing an explicit cmake.h include. Some were not including
it at all. Now it's used almost everywhere.
(cherry picked from commit 82ae86979c497e6d1d0c6b2b5a55aa379ec82c98)
- File format 2 (used in version 0.9.3 - 1.5.0) is no longer supported.
- Task::parseJSON implemented.
- PRODUCT_TASKWARRIOR defined, to facilitate code sharing.
- Strings renamed for sharing purposes.
- When taskwarrior is built without GnuTLS, there will be a message to that
effect.
- Stubbed the TLSClient calling code. The current Socket code will become
obsolete soon.
- Cmake was not updating HAVE_ST_BIRTHTIME.
- NIBBLER_FEATURE_DATE was not properly applied everywhere.
- FEATURE_COLOR was not properly set.
- Some source files failed to include cmake.h, and therefore were not properly
- Removed inefficient use of std::string::substr for guaranteed single character
strings.
- Integrated Directory::cd.
- Integrated File::ctime, ::btime.
- Integrated Path::operator+.
- Integrated Nibbler::getDigit{2,4,6}.
- Integrated HighResTimer.
enabling/disabling code.
- All Path objects now expanded internally to absolute form.
- Modified unit tests to accomodate absolute paths.
- Merged new nibbler.t.cpp tests.
- Made various methods const.
- Includes removed from some files, added to others.
- Modifed the color.duetoday.t test to look for 'red', which for some reason
fails on Ubuntu 12.10, *sometimes*. This change removes the red/rgb400
duality, which was wrong anyway. We'll call this 'cleanup'.
- Any test that contains a '*' in the description output has a chance of failing
because that '*' gets expanded by the shell, somewhere in the run_all script.
While I can't find that problem (yet) and fix it, I can remove the '*' from
the output.
