caches are module global in the zip_directory_cache. When it is
updated due to a changed zip file, all zipimporter instances need to
see the same updates TOC cache. This fixes the bug for the overlooked
submodule import case from the earlier round of changes. Includes
tests that would fail otherwise.
It also refactors zipimporter_init in the process to make it a bit
easier to read and understand. Less reuse of the same variable for
multiple purposes and the local path buffer is malloc'ed instead
of consuming a large MAXPATHLEN+2 chunk stack space.
from is modified during the lifetime of the Python process after
zipimport has already opened and cached the zip's table of contents
it now fstat's the file after opening it upon every attempt to access
anything within and will re-read the table of contents if the .zip file
inode, size or mtime have changed.
It would've been nicer to hold any .zip file used by zipimport open for the
duration of the process but that would be more invasive and add an additional
open file descriptor to all zipimport using processes. It also would likely
not fix the problem on Windows due to different filesystem semantics.
var_decl: Declared variable "stm" without initializer
ninit_use_in_call: Using uninitialized value "stm" (field "stm".tm_zone uninitialized) in call to function "mktime"
using a custom, nearly-identical macro. This probably changes how some of
these functions are compiled, which may result in fractionally slower (or
faster) execution. Considering the nature of traversal, visiting much of the
address space in unpredictable patterns, I'd argue the code readability and
maintainability is well worth it ;P
