I should have left the entire context in my reply.

> On May 31, 2016, at 1:59 PM, Xiaodi Wu <[email protected]> wrote:
> 
> On Tue, May 31, 2016 at 2:51 PM, Christopher Kornher via swift-evolution 
> <[email protected] <mailto:[email protected]>> wrote:
> 
>> On May 31, 2016, at 1:47 PM, Xiaodi Wu <[email protected] 
>> <mailto:[email protected]>> wrote:
>> 
>> 
>> 
>> On Tue, May 31, 2016 at 2:45 PM, Christopher Kornher via swift-evolution 
>> <[email protected] <mailto:[email protected]>> wrote:
>>> 
>>> Not allowed:
>>> …
>>> let a = a
>>> let b = b where b > 10 && a > 5
>>> 
>>> Why would this not be allowed by your rule? You're making use of `b` in 
>>> your where clause. As I demonstrated above, essentially any assertion can 
>>> be rewritten to work around your rule. In general:
>> 
>> It is not allowed because  ‘a’ is defined in the line above. It must be 
>> defined in the ‘if let’ associated with the where in which it is mentioned.
>> 
>> That's a much more restrictive where clause than you proposed earlier. You'd 
>> not be able to write:
>> 
>> ```
>> let b = b where b > anyOtherVariable
>> ```
> 
> 
> The definition is not a formal one, but that was the intent.
> 
> ```
> let b = b where b > anyOtherVariable
> ```
> is legal as long as `anyOtherVariable` is not defined within the entire 
> condition clause
> 
> 
> You can propose that rule, but it doesn't solve the issue. If, today, I've got
> 
> ```
> let x = 1
> let y: Int? = 2
> let z = 3
> 
> if let y = y where x < z {
>   // do stuff
> }
> ```
> 
> your rule simply forces
> 
> ```
> if let y = y where y == y && x < z {
>   // do stuff
> }
> ```
> 
> The point is, the semantic relationship between what comes before and after 
> `where` exists in the mind of the human reader only.

I meant to add that all boolean expressions after the where must use one of the 
constants defined in the associated `if let` so the `&& x < z` would not be 
allowed.

I don’t understand the 'y == y’ in your example


_______________________________________________
swift-evolution mailing list
[email protected]
https://lists.swift.org/mailman/listinfo/swift-evolution

Reply via email to