[swift-evolution] SE-0025: Scoped Access Level, next steps

[Ross O'Brien]

So, just so I'm understanding correctly what the new system would be
(taking into account Chris Lattner's comments).

A symbol specified as 'public' is visible outside the module.
A symbol with no specifier, or specified as 'moduleprivate', is visible
only within the module, but anywhere within it. This default lets us start
coding first and refine visibility later; 'moduleprivate' replaces
'internal'.
A symbol specified as 'fileprivate' is visible only within the file in
which it is declared.
A symbol specified as 'private' is *as private as possible* given its type.

So, if a constant, variable, property, func, enum, struct, etc. is defined
inside of a scope, 'private' makes it visible only within that scope. All
private properties of a nested inner type are visible throughout the scope
of the outer type which contains it, and the inner type can see any private
properties in the scope which contains it.

If a constant, variable, protocol, class etc. is defined outside of a
scope, 'private' is synonymous with 'fileprivate'. (I don't know if using
'fileprivate' on a type which cannot be "scope-private" should generate a
warning.)

Extending to the future: if we add any more levels, they must fit within
the hierarchy (i.e. all levels must be expressible as concentric rings in a
Venn Diagram, no overlapping) and there must be an objective specifier
which signifies its visibility.
So if we add a 'submodule' level, then a submodule must comprise a set of
whole files which is a subset of the containing module. A
'submoduleprivate' symbol is visible anywhere in any of the files it
consists of - there would be no way to mark part of a file as being within
the submodule, and part of the file outside of it.


On Wed, Mar 30, 2016 at 2:33 PM, David Waite via swift-evolution <
swift-evolution at swift.org> wrote:

> I believe we will need another non-type access level if frameworks are
> ever divided into submodules.
>
> -DW
>
> On Mar 30, 2016, at 7:13 AM, Ilya Belenkiy via swift-evolution <
> swift-evolution at swift.org> wrote:
>
> I am not sure if we will ever get another access level. If we do, great,
> but given how long this discussion has been already, I am not counting on
> it :-)
>
> Most likely, if we get more, it will be possible to describe it with a
> simple word, or a combination of words or with some common abbreviations,
> so I am not worried about extensibility. I think that the names in the
> proposal are very consistent with Swift as it is today and will serve us
> well. They are also completely unambiguous and don't depend on the reading
> context, so if we come up with other ways to label access levels, it should
> still be possible to either use these names for backward compatibility or
> migrate them automatically to new names without any difference in semantics.
>
> We also needed to pick something. I waited for about a week to get
> everybody's vote, and I think that I picked a compromise that we can all be
> at least ok with. (I also originally wanted short single word names). I
> think we should close the naming thread at this point.
> On Wed, Mar 30, 2016 at 5:26 AM Matthew Judge <matthew.judge at gmail.com>
> wrote:
>
>>
>>
>> On Mar 29, 2016, at 20:47, Brent Royal-Gordon via swift-evolution <
>> swift-evolution at swift.org> wrote:
>>
>> >> If Scala style access modifiers were adopted for Swift then a
>> private(file) modifier would also be necessary to give the current private
>> functionality.
>> >
>> > I could imagine having these options:
>> >
>> >    public                            // visible to all everyone
>> >    private(scope-name, scope-name, …)    // visible to specified scopes
>> (plus current scope)
>> >    private                            // visible only to current scope
>> >
>>
>> Allowing multiple "scope-name"s is a lot of flexibility and power, but
>> not sure it's useful/worthwhile.
>>
>>  For the current discussion, I would think "scope-name" should be limited
>> to an enclosing scope only.  So you can say "private(Outer)" from an Inner
>> class or "private(#file)" from within a class, but not "private(ClassA)"
>> from within ClassB.
>>
>> (This would also solve the ambiguity of how to reference the main ClassA
>> or a specific extension to ClassA... "private(ClassA)" can only refer to
>> whichever scope of ClassA you are currently in.)
>>
>> > scope-name could perhaps be:
>> >
>> > * A type name (or Self, which would mimic C++-style private, or perhaps
>> even C++-style protected depending on how we treat inheritance)
>>
>> But, this is getting into type-based access which is beyond the scope of
>> SE-0025 right?
>>
>> > * A module name (or #module for the current module)
>> > * A file name string (or #file for the current file)
>> >
>> > And then the default would simply be `private(#module)`.
>> >
>> > Alternatively, the parameterized level could be given a different name,
>> like `internal` or `shared`. If that were the case, then `#module` might
>> simply be the default.
>> >
>> > --
>> > Brent Royal-Gordon
>> > Architechies
>> >
>> > _______________________________________________
>> > swift-evolution mailing list
>> > swift-evolution at swift.org
>> > https://lists.swift.org/mailman/listinfo/swift-evolution
>>
> _______________________________________________
> swift-evolution mailing list
> swift-evolution at swift.org
> https://lists.swift.org/mailman/listinfo/swift-evolution
>
>
>
> _______________________________________________
> swift-evolution mailing list
> swift-evolution at swift.org
> https://lists.swift.org/mailman/listinfo/swift-evolution
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.swift.org/pipermail/swift-evolution/attachments/20160330/1e19ca3c/attachment.html>