[swift-users] Compile time exceeded. Anything wrong?

[Geordie Jay]

I doubt it'll be "fixed" in the next release (which came out less than 24
hours ago), but I gather type checker speed optimisations and fixing
crashes are ongoing goals of the compiler team. With no guarantees or
expectations, you may find your code compiles in an acceptable time with
Swift 4 (Xcode 9).

Cheers,
Geordie
On Tue 6. Jun 2017 at 12:29, <www.hbu.cn at 163.com> wrote:

> thanks,  guys!  dp array should be var, not let, but the compile run too
> slow. It indeed a bug.  Optimizing  the code for the compiling time is
> really a headache for coders.  I am using the Xcode 8.1. Hoping it can be
> fixed in next release.
>
>
> *best wishes for you *
>
> 2017-06-06 8:57 GMT+08:00 Geordie J <geojay at gmail.com>:
>
>>
>> Am 06.06.2017 um 09:02 schrieb Jens Persson via swift-users <
>> swift-users at swift.org>:
>>
>> When compiling that from the command line, I get the following (after
>> about 6 seconds):
>>
>> test.swift:7:18: error: cannot assign through subscript: 'dp' is a 'let'
>> constant
>>         dp[0][0] = 0
>>         ~~       ^
>> /.../
>>
>> After fixing that (changing let dp to var dp), it will compile
>> successfully, still taking a very long time though. This usually means that
>> some expression(s) in the code happen to be a bit hard on the type checker
>> (in its current state).
>> I tried the latest dev snapshot and it is a bit faster, perhaps 3 s
>> instead of 6.
>>
>> Anyway, here is a logically equivalent rewrite which will compile faster:
>>
>> class Solution {
>>     func rob(nums: [Int]) -> Int {
>>         guard nums.count > 0 else { return 0 }
>>         var dp = Array.init(repeating: Array.init(repeating: 0, count:
>> nums.count),
>>                             count: 2)
>>         dp[0][0] = 0
>>         dp[0][1] = nums[0]
>>         for i in 1 ..< nums.count {
>>           let dp_iMinus1_0 = dp[i - 1][0]
>>           let dp_iMinus1_1 = dp[i - 1][1]
>>             dp[i][0] = max(dp_iMinus1_0, dp_iMinus1_1)
>>             dp[i][1] = dp_iMinus1_0 + nums[i]
>>
>>
>> Just a nitpick: this isn’t functionally equivalent to the original code
>> (dp[i][1] = dp[i][0] + nums[i]), because dp[i][0] might have actually
>> changed on the previous line. But you can return this line to the original
>> version without any effect on the compile time (less than 250ms on my
>> machine).
>>
>> I think the compile time issue is that max() is a generic function that
>> takes any Comparable, so the type checker seems to go berserk trying to
>> ensure the term is satisfiable.
>>
>> If you create a non-generic replacement for max in the same file:
>>
>> private func myMax(_ x1: Int, _ x2: Int) -> Int {
>>     return (x1 >= x2) ? x1 : x2
>> }
>>
>> … and replace the assignment to *max(dp[i - 1][0], ……)* in the for loop
>> with *myMax(dp[i - 1][0], ….)*
>>
>> The compile time will be just as fast as the one that swaps out the
>> internal elements (on my machine it’s actually about 10% faster with the
>> non-generic *max*).
>>
>> Regards,
>> Geordie
>>
>>         }
>>         return 0
>>     }
>> }
>>
>>
>>
>>
>>
>> On Mon, Jun 5, 2017 at 2:32 PM, Hbucius Smith via swift-users <
>> swift-users at swift.org> wrote:
>>
>>> Hi Swift-Users,
>>>
>>>     when I compiled the code, Xcode cannot stop, I do not know why. It
>>> is very strange. Can anyone help ? Here is the code. I am using Xcode 8.1
>>>
>>> class Solution {
>>>
>>>     func rob(nums: [Int]) -> Int {
>>>
>>>         guard nums.count > 0 else { return 0 }
>>>
>>>         let dp = Array.init(repeating: Array.init(repeating: 0, count:
>>> nums.count),
>>>
>>>                             count: 2)
>>>
>>>         dp[0][0] = 0
>>>
>>>         dp[0][1] = nums[0]
>>>
>>>         for i in 1 ..< nums.count {
>>>
>>>             dp[i][0] = max(dp[i - 1][0], dp[i - 1][1])
>>>
>>>             dp[i][1] = dp[i - 1][0] + nums[i]
>>>
>>>         }
>>>
>>>         return 0
>>>
>>>     }
>>>
>>> }
>>>
>>>
>>>
>>> *best wishes for you *
>>>
>>> _______________________________________________
>>> swift-users mailing list
>>> swift-users at swift.org
>>> https://lists.swift.org/mailman/listinfo/swift-users
>>>
>>>
>> _______________________________________________
>> swift-users mailing list
>> swift-users at swift.org
>> https://lists.swift.org/mailman/listinfo/swift-users
>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.swift.org/pipermail/swift-users/attachments/20170606/cc514b3f/attachment.html>