And I want to remind about my typical opinion: We should try to require
as little extensions as possible and do as many things as possible (in a
usable way) in userland. If we spend time on making PHP faster we reduce
the need for extensions even more.
You're assuming the target implementation is trivial, or even possible. In non-trivial libraries/implementations, you'll find developers chose to write their library in c/c++ b/c c and c++ are portable.
Off the top of my head, an example of this would be mapnik. Currently, no good extension wrapping libmapnik exists, nor is there any spec to follow that would allow a php developer to replicate the usefulness of that library in userland php. Moreover, if someone were to be able to create a userland/1-1 implementation of it, they'd be hard pressed to keep up with that projects momentum in improving the baseline implementation in c++.
Another example is the AOP project. Without an extension hijacking the executor, this kind of extension is not possible in userland in any meaningful (1st class) kind of way.
I could probably come up with a few more examples, but I think those suffice.
More to the point though- having the ability to expose portable libraries to PHP though extensions underscores the "glue" aspect of PHP's usefulness, it would be beneficial (IMO) to expand this usefulness to developers.
-ralph