Configuration docs refresh
- Moved the technical details of the config system into the development directory.
- Added a new 'intro' document describing how to use the config system, without the technical info.
- Autogenerate lists of config options for the main components (this doesn't yet include the parallel machinery).
- Condense and tidy up other config docs, removing outdated information.
Daemon StreamCapturer
The StreamCapturer should die if the main thread crashes. On Shiningpanda, a failure in another nose plugin has been causing the tests to hang, because the main thread exits, but the StreamCapturer thread is still alive.
Under normal conditions, the thread will still be shut down cleanly - it will only die a messy death if the main thread does.
various pandoc latex fixes
- add missing declarations to latex_base (required for pandoc-highlighted code in markdown)
- fix ansi color in latex, which didn't work in a variety of ways.
closes#4245
update script generation in setup.py
These changes:
- Give us scripts called `ipython` and (`ipython2` or `ipython3`) for any installation. (and likewise for each of iptest, ipcontroller, etc.)
- Add a new `setup.py symlink` target, to use instead of `develop`, which installs scripts and symlinks the library into site-packages, without using setuptools.
- Removes the static script entry points - all our entry points are now automatically generated.
Python 3 & getcwdu
We were using the os.getcwdu() function in several places. That doesn't exist on Python 3, but the path.py module was masking that by setting os.getcwdu = os.getcwd. As described in #4462, however, that didn't always work.
This adds a reference in py3compat which points to getcwd on Python 3 and getcwdu in Python 2.
a lunch conversation with @minrk and @takluyver lead me to just make a
popup and have it be closed, instead of having to reorder the test suite
so that this test always runs last. Now it doesn't matter what order the
test suite runs in, and gives us a pattern for testing window closing
behavior should we need it elsewhere.
Any content whose metadata contains an `isolated` tag will be isolated
from the rest of the document.
The current implementation wraps isolated content into an iframe.
