Kenneth McDonald wrote: > Why would any of the issues below be any more difficult than they are with > the current file functions? I'm not proposing a C replacement for current > functions, merely a Python module that wraps all of those functions (and > adds some additional ones) in an appropriate class.
I'm not saying they are difficult. I want to know how your library deals with them. There is a good chance that some of these questions remain unanswered in the PEP, and I just want to indicate that I would be unhappy if they are. Specifying this API is a huge task, much more so than coming up with an implementation that does "something". This is one of the reasons why nothing like this has made it to the standard library: as a library module, it would have to face many more scenarios that the authors of the module originally did not consider. Therefore, the documentation must be complete and consistent, and there should be an agreement as to what this library can do and what it cannot do. Regards, Martin -- http://mail.python.org/mailman/listinfo/python-list
