Author jvr
Recipients
Date 2002-12-14.20:46:01
SpamBayes Score
Marked as misclassified
Message-id
In-reply-to
Content
Logged In: YES 
user_id=92689

Btw. it would be nicer if I indeed provided an "importer"
wrapper for the builtin mechanism and simply placed it on
and invoke it through sys.meta_path. The reason I haven't
done that yet is that A) this would change import.c even
more (although it would be better factored then) and B) it
would have _some_ impact on the performance of the builtin
import mechanism, as *all* imports would then pass through
yet another Python call. But I hope to play with this a bit
later next week. It could also be a project for 2.4.

Note that this is exactly what Gordon does in iu.py, except
he goes even further: he has separate importer objects on
the metapath for builtin modules, frozen modules and file
system imports. I think that's fantastic, but it would be
more controversial as it might affect performance and might
be harder to integrate with the current imp module.
History
Date User Action Args
2007-08-23 15:18:57adminlinkissue652586 messages
2007-08-23 15:18:57admincreate