X-Comment1: #############################################################
X-Comment2: # uk.ac.glasgow.cs has changed to uk.ac.glasgow.dcs #
X-Comment3: # If this address does not work please ask your mail #
X-Comment4: # administrator to update your NRS & mailer tables. #
X-Comment5: #############################################################
(As you'll have gathered, I'm trying to tie up all the changes to the
report required for a March appearance in SIGPLAN Notices. That means
finishing by Xmas; hence my recent activity!)
I have two matters concerning import/export/hiding lists I'd like to raise.
Simon
Synonyms
~~~~~~~~
At present a type synonym can be exported by saying
module A(T)
type T = Int
The type T is not abstract, because the full synonym is exported. If T
were declared with a data decl, then T would become abstract. This seems
odd. Mark Jones suggests requiring the form T(..) for exporting or importing
type synonyms, for consistency's sake. (It also allows a possible future
form of abstraction by exporting a synonym without its representation.)
Under this proposal the form T would be illegal for type synonyms (T(..)
is currently illegal).
This proposal interacts slightly with the one below.
Hiding
~~~~~~
Lennart asks:
| If I have an interface
|
| interface A where
| data T = X | Y
|
| and then do the following
|
| module B where
| import A hiding(T)
| ...
|
| Does that mean that only the type is hidden,
| but not the constructors (X, Y)?
| Or is it the same as
|
| module B where
| import A hiding(T(..))
| ...
This seems like a Jolly Good Question. It is obviously silly to hide the
type without its constructors.
Proposal: in hiding lists allow only the forms
var
T(..) -- Types
C(..) -- Classes
If we don't adopt the earlier proposal for synonyms we'd need an extra form
T for type synonymns.
Simon
