Should PyPDF2 2.0.0 expose everything directly under PyPDF2?

[MartinThoma]

Currently, PyPDF2 has the following public modules:

• errors: All Exceptions
• generic: NullObject, BooleanObject, ArrayObject, DictionaryObject, IndirectObject, ...
• utils: readUntilRegex, skipOverComment, skipOverWhitespace, ...
• pdf: will be removed with PyPDF2 2.0.0
• merger: will become private to be consistant with _reader.py and _writer.py.
• pagerange: PageRange, parse_filename_page_ranges

Is there any reason NOT to expose all public classes / functions under PyPDF2 (via __all__) directly?

The advantage of exposing it there would be:

1. The users don't have to memorize how PyPDF2 is internally structured
2. PyPDF2 can be restructured easily as the public interface is clearly defined in one place

[pubpub-zz]

I agree about errors/generic/pagerange/merger
I have some doubts about utils : unless I'm wrong they are only dealing with internal functions,
Also what could be a good invitation to use the new syntax could be to make public the new names (PdfReader) but not the old one (PdfFileReader)