At 11:01 AM 8/4/2001 -0400, Geoff Talvola wrote:
>At 11:41 PM 8/3/2001 -0400, you wrote:
>
><snip>
>
>>In the process of fixing the AbsolutePath directory index problem, I
>>realized that AbsolutePath requests weren't going to work with our
>>latest code because they don't have a context, and one cannot be created
>>for them by creating a file named __init__.py next to them(due to
>>permission issues). So I added some code to ServletFactory to create
>>packages for these file in a new way that doesn't require creating an
>>__init__.py file. It seems to work correctly, but Geoff, please take a
>>look at it.
>
>I should point out that I've never tried the AbsolutePath feature -- I
>always use a regular Webware context.
>
>I think this is the code you're talking about:
>
> ## There is no context, so create a package name that is the same
> as the filesystem path
> if fullname == None:
> remainder = path
> remainder = string.replace(remainder, '\\', '/')
> remainder = string.split(remainder, '/')
> moduleName = string.join(remainder,".")
> if moduleName[0] == ".":
> moduleName = moduleName[1:]
> directory, contextDirName =
> os.path.split(serverSidePathToImport)
> if debug: print moduleName, contextDirName, directory
> filename = contextDirName[:string.rfind(contextDirName, '.')]
> module = self._importModuleFromDirectory(moduleName,
> filename, directory)
> return module
>
>If I understand the code correctly, it's creating a "fake" package name
>based on the filesystem path? So if the file is located at
>/foo/bar/qux/mypage.py it'll import the module as foo.bar.qux.mypage? If
>that's right, it sounds safe to me.
>
>You're not really following the normal rules for importing a package,
>which are listed in the documentation for the imp standard library module
>-- here's a snippet of the documentation for imp.find_module():
>
>>This function does not handle hierarchical module names (names containing
>>dots). In order to find P.M, i.e., submodule M of package P, use
>>find_module() and load_module() to find and load package P, and then use
>>find_module() with the path argument set to P.__path__. When P itself has
>>a dotted name, apply this recipe recursively.
>
>In other words, first import the package itself (i.e. the __init__.py) and
>then the module within the package. These are the rules that the context
>importing code follows, and that's why you need an __init__.py in each
>context directory and subdirectory.
>
>I wonder if we can relax the requirement for an __init__.py in each
>subdirectory of a context? I'll do a bit of experimenting. I was
>reluctant to deviate from the standard Python package importing rules but
>if this is working for the absolute path case, maybe it'll work for
>context imports also.
I tried an experiment -- in ServletFactory._importModuleFromDirectory() I
removed the code that writes out an empty __init__.py if it's
missing. Instead, I just had it skip the import of the package if
__init__.py was missing.
This failed and gave me numerous strange errors about my servlets not being
derived from Servlet.
From this experiment I conclude that you really _do_ need to have an
actual __init__.py if you're going to import something as a package,
otherwise something goes wrong in the import mechanism. So I think the
code snippet above could be dangerous.
I remember many months ago when I was writing the "import contexts as
packages" code I tried for a long time to find a way to fake it and not
require the __init__.py file to be on disk, but I could never get it to work.
- Geoff
_______________________________________________
Webware-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/webware-devel
