All python docs and description indicate that optimization (-OO) does not do
much anything except the removal off pydoc. A single "O" removes comments and
asserts, and with the removal of pydoc with double "O" option the *.pyo byte
compile is left with pure executable code. I am experiencing a different
behavior than described.
I am running Python 2.6.4 and have source code which I pre-compile either into
pyc or pyo files depending on the optimization switch selected. The pyo
version fails to run with the main program module failing to import any other
modules, such as failing on the "import os" statement (first line
encountered). However, the pyc version succeeds and runs correctly. This is
with the same code modules, same python VM and same machine.
One item I should note is that the Python distribution I am using is not fully
installed with paths set by the installer. I unpack the Python tar and compile
it (i.e. ran configure and make; not make install). Then I distribute the this
Python VM, with its Lib and Modules dirs into my target machine (a Linux
distro) and collocate my pyc or pyo modules at the root with python.
To further experiment, I have also compiled all python libraries to either pyc
and pyo (e.g. os.pyc or os.pyo in the Modules dir). If I then run with
interactive python, I experience the same effect as executing from command
line. Under the IDE "import os" fails if I distribute with python modules
compiled into pyo, but it succeeds if I distribute pyc modules.
This seems to be contrary to the documentation. If the only difference was the
removal of pydoc between pyc and pyo then both versions should behave exactly
the same way. There must be some additional modifications with a -OO compile.
Anyone can comment, please?
Thanks,
Boris
--
http://mail.python.org/mailman/listinfo/python-list
