Are old assumptions still valid?

[luser- -droog]

In the manpage from Crispin Goswell's postscript
interpreter, he mentions a number of decisions about
the correct behavior of postscript that he was forced
to make without guidance from the red book.

Are these assumptions still valid in the level2/level3
world?

ASSUMPTIONS
There are a number of Assumptions that have been made in the
absence of more specific documentation.

That Integer and Real values are only equivalent mathe-
matically, and are normally distinct except where speci-
fied in the manual. The manual requires equivalence in
the following places:

The arguments of the arithmetic and mathematical
operators.

The arguments of the relational operators
(including eq and ne).

In particular integers and reals are distinguished as
dictionary keys.

That cvrs is not expected to convert reals to any base.

That the output from == and pstack can look however we
want them to.

That %stdout and %stderr should not ever be closed - the
manual is specific about %stdin, but not the other two.

That stop can exit a run context.

That flushfile should close an input stream. Read will
close its file if it reads nothing, but the manual
doesn't say whether other operators should. It is
assumed so.

That access restrictions on files are implemented using
the access restrictions on objects.