make the core docbuild function even simpler; have it determine the result
and return it, as well as the individual build success/failure vector
move all logging logic into the caller function, so that script(), publish()
and build() can log whatever they like
adjust calling pattern for prepare_{script,build}_mode so that they are easier
to test
embed the creation of build directories into the prepare_build_mode
remove the logic from the docbuild function which executes both the --script
generation and the --build generation as the core loop over the document set.
create ancillary functions to prepare the document set for --script mode or
--build mode
add functions to create and remove the build directories in --build mode
add a --publish cleanup function to leave our --builddir clean (unless there
was a failure)
created function removeOrphans() and removeUnknownDoctypes()
and the function runbuild() ugly name; which is called from all three of the
main work functions, build(), publish() and script()
abstract the error-handling away from the one-large run() function into each
of the functions (show_doctypes, show_statustypes, detail, summary, etc.)
add the function publish(), which will call build() and ensure success before
running any of the publishing
In preparation for supporting a separate --builddir (allowing minimal
disruption of real output directory during rebuild) factor all output
directory handling logic into the main processor object (BaseDoctype).
Simplify the generate() method.
Centralize all pre-build logic in hook_build_prepare().
Remove all hook logic from the OutputDirectory.
the logic for making sure to chdir() into the build directory has been
sequestered into doctypes/common.py (and output.py); additionally, it is
smarter to put the resource copying logic, there, as well
in preparation for supporting a build-directory, moving the os.chdir() and the
copying of image files into tldp/doctypes/common.py and adding a few hooks in
the main logic for building