[swift-evolution] [Discussion] A Problem With SE-0025?

Matthew Johnson matthew at anandabits.com
Wed Jun 29 11:10:30 CDT 2016


> On Jun 29, 2016, at 11:06 AM, Xiaodi Wu <xiaodi.wu at gmail.com> wrote:
> 
> On Wed, Jun 29, 2016 at 11:03 AM, Matthew Johnson <matthew at anandabits.com <mailto:matthew at anandabits.com>> wrote:
> 
>> On Jun 29, 2016, at 10:55 AM, Xiaodi Wu <xiaodi.wu at gmail.com <mailto:xiaodi.wu at gmail.com>> wrote:
>> 
>> On Wed, Jun 29, 2016 at 10:52 AM, Matthew Johnson via swift-evolution <swift-evolution at swift.org <mailto:swift-evolution at swift.org>>wrote:
>> 
>>> On Jun 29, 2016, at 10:46 AM, Sean Heber <sean at fifthace.com <mailto:sean at fifthace.com>> wrote:
>>> 
>>>> 
>>>> On Jun 29, 2016, at 10:22 AM, Matthew Johnson via swift-evolution <swift-evolution at swift.org <mailto:swift-evolution at swift.org>> wrote:
>>>> 
>>>> 
>>>>> On Jun 29, 2016, at 10:08 AM, David Hart <david at hartbit.com <mailto:david at hartbit.com>> wrote:
>>>>> 
>>>>> Sorry if I wasn’t expressing myself well enough. In my original email, I said that:
>>>>> 
>>>>>> The new rules make `private` more prominent compared to `fileprivate` (the latter has a somewhat worse name).
>>>>> 
>>>>> So I agree that my issue is more with the naming than the functionality. I’m mainly complaining that because of its name, `fileprivate` feels like more of a special corner case of `private`. But in the style of writing types through extensions, `fileprivate` will become much more prevalent than `private`, which feels slightly backwards.
>>>> 
>>>> I don’t view it as more of a special corner case at all, but I can see why you feel that way since it has an unprecedented (AFAIK) and more verbose name.  
>>>> 
>>>> The proposal originally left `private` alone and used a new name for the new access level.  We weren’t able to find a name that didn’t have problems which led to the idea of renaming the existing `private`.
>>>> 
>>>> My perspective is that it’s just the best name we could come up with for the concept in the context of the various access levels we want to support.  The name isn’t intended to discourage use in any way.  
>>> 
>>> It may not be intended, but that doesn’t mean it won’t, though. :P
>>> 
>>> I can’t say exactly *why*, but I feel similar to David here in that “fileprivate” is such an… odd… name that I’m inclined to just not use it and let things default to “internal” instead. In fact, I have *already* caught myself doing this. I don’t know if that’s *bad* exactly (would more things being internal actually aid the compiler/optimizer?), 
>> 
>> I’m pretty sure more things being internal will not help the optimizer.  In fact, if you are not compiling with WMO turned on it could prevent optimizations.
>> 
>>> but I think this is a valid concern. The issue here is rooted in psychology, not technology. :/
>> 
>> That’s a fair perspective.  But a *significant* amount of time was spent bike shedding this.  I’m not sure whether you and David participated or not, but that was the time to have the naming discussion.
>> 
>> I think the case being made here is that `fileprivate` was settled on when it was thought that it would be rarely used. With what's emerged in this discussion, it turns out that `fileprivate` might be more useful than previously thought, and the awkwardness of the name therefore is more troublesome than when the naming discussion first took place.
> 
> The example in this thread (placing data members in the type declaration and methods in extensions) is one that received ample discussion during the earlier threads and the review.
> 
> I don’t know that `fileprivate` will be used in code more commonly than previously thought.  The issue is about the default access level of members inside a `private` type (i.e. when access is *not* directly specified).  With Jordan’s proposed solution, `fileprivate` will be used to describe these members in documentation and diagnostics.  
> 
> It will also be possible to state this default explicitly, but I don’t think that will be too common.  This is the only change in what is possible to do *in code* from the original proposal.
> 
> You're adding words to my argument that I didn't put there. I didn't specify "in code". Awkward is awkward, in code or in documentation.

That wasn’t intentional, sorry.  I misunderstood and thought you meant it would be used more frequently in code than previously thought.  Thanks for clarifying.

I certainly don’t oppose a better name if anyone can suggest one that is clearly better.  But I am skeptical that this is possible given the amount of bike shedding that has already happened.

>  
> 
>> 
>> 
>> IMO the value of having more control over visibility far outweighs a slightly awkward name for file level visibility.  I don’t think it’s anywhere near awkward enough to avoid, but I suppose YMMV.
>> 
>>> 
>>> l8r
>>> Sean
>> 
>> 
>> _______________________________________________
>> swift-evolution mailing list
>> swift-evolution at swift.org <mailto:swift-evolution at swift.org>
>> https://lists.swift.org/mailman/listinfo/swift-evolution <https://lists.swift.org/mailman/listinfo/swift-evolution>
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.swift.org/pipermail/swift-evolution/attachments/20160629/fe6777d8/attachment.html>


More information about the swift-evolution mailing list