A smarter(?) package importer.

[Prabhu Ramachandran]

>>>>> "GMcM" == Gordon McMillan <gmcm at hypernet.com> writes:

    >>  Well, I'm not sure it does 3/4 opens but yes it does do 3/4
    >> imp.find_modules which will be more expensive than a
    >> sys.modules[mod_name].

    GMcM> It really does do all those opens. If Windows had strace, it
    GMcM> would be easy to verify.

Yes, you are right.  find_module does use open underneath.  Thanks for
the information.

write(1, "Importing foo (pkg/sub1/foo.pyc)"..., 37Importing foo (pkg/sub1/foo.pyc)... 
) = 37
stat("pkg/sub1/a", 0xbfffe08c)          = -1 ENOENT (No such file or directory)
open("pkg/sub1/a.so", O_RDONLY)         = -1 ENOENT (No such file or directory)
open("pkg/sub1/amodule.so", O_RDONLY)   = -1 ENOENT (No such file or directory)
open("pkg/sub1/a.py", O_RDONLY)         = -1 ENOENT (No such file or directory)
[snip]
 
    GMcM> [back to recursive-relative imports]

    >> So, I'll ask again, how about making this behaviour standard?
    >> :)

    GMcM> I think not. I'd rather see import get less magical than
    GMcM> more magical.

    GMcM> If "import" meant "absolute import", and "rimport" meant
    GMcM> "relative import" (and maybe "rrimport" meant "recursive
    GMcM> relative import") then all name- masking problems would
    GMcM> disappear.

That sounds ok but this adds 2 more keywords to Python.  Anyway, so
long as something that handles rr/rimport is added I'm happy.  Should
I make a request on python-dev about this?

prabhu