[IronPython] Some new bugs with asssembly loading...
xtian at babbageclunk.com
Thu Jan 5 11:13:22 CET 2006
I don't think that would work very well - assemblies and namespaces
are orthogonal. The Python equivalent to adding a reference to an
assembly is adding a path (or an egg or zip) to sys.path.
That's important, because an assembly can contain many different
namespaces (in the same way an egg can contain multiple packages or
modules, although it's not quite equivalent). You wouldn't necessarily
want all of the classes in an assembly to be imported into the global
The main difference is that if you have two assemblies that expose the
same namespace, when you do a using statement for that namespace,
you'll have all of the members of the namespaces in both assemblies
available, while in Python, you'd get only the members of the module
that was first on the path. There are pluses and minuses for both
On 1/5/06, Keith J. Farmer <kfarmer at thuban.org> wrote:
> Actually, I'm wondering about the usefulness of an assembly delimiter for the import statement.
> IE, instead of clr.Add*, you have something more like:
> from (FooAssembly, Version=18.104.22.168, Culture=neutral, PublicKeyToken=b77a5c561934e089) import *
> This maintains the from/import structure in current use -- it just expands the type that the first argument can handle to include, instead of a python library identifier, an assembly reference. It'd be easier to parse, I think, when dealing with the whole IronPython->Assembly path, rather than hunting for scattered clr.Add* references.
> From: users-bounces at lists.ironpython.com on behalf of Martin Maly
> Sent: Wed 1/4/2006 9:41 AM
> To: Discussion of IronPython
> Subject: Re: [IronPython] Some new bugs with asssembly loading...
> The reason is that clr.References is a tuple without improved formatting. Using tuple was a way to get the idea out there and gather feedback and experiences. Ultimately we may use some custom written tuple-like structure.
> users mailing list
> users at lists.ironpython.com
More information about the Ironpython-users