Sort Method for 5 var's type Integer

[engkin]

Nothing like being lazy.

I totally agree. Have a look at this:

Code: Pascal  [Select][+][-]

1. program sortarray;
2. {$mode delphi}{$MACRO ON}{$IF FPC_FULLVERSION < 30101}{$ERROR NEEDS FPC 3.1.1 or HIGHER}{$ENDIF}
3. uses generics.collections;
4. var
5.   a:TArray<integer>;
6.   i:integer;
7. begin
8.   a :=[3,1,2,5,4];
9.   TArrayHelper<integer>.Sort(a);  // NOTE: This is a class. Not a type helper. Will be renamed/added to Tarray<T> in a future version.
10.   for i in a do writeln(i);
11. end.

That's all (in trunk...) Quite a bit lazier, I guess...

Lazier is definitely better.

[engkin]

But that takes a separate download? Anyway you did a nice job!

Yes. I was in opposition to including rtl-generics into 3.0.4 (it was considered). So thanks to me we don't have rtl-generics in 3.0.4 and you need to download library on your own .

Now we know whom to thank for that crime.

[engkin]

Test for "Nothing like being lazy"

Enter size of array (<=0 to end): 5
Enter max integer: 1000
Enter test count: 1000
SortBits elapsed: 31
SortGeneric elapsed: 16
--------------------
Enter size of array (<=0 to end): 50
Enter max integer: 1000
Enter test count: 1000
SortBits elapsed: 31
SortGeneric elapsed: 16
--------------------
Enter size of array (<=0 to end): 500
Enter max integer: 1000
Enter test count: 1000
SortBits elapsed: 47
SortGeneric elapsed: 171
--------------------
Enter size of array (<=0 to end): 5000
Enter max integer: 1000
Enter test count: 1000
SortBits elapsed: 94
SortGeneric elapsed: 1217
--------------------
Enter size of array (<=0 to end): 50000
Enter max integer: 1000
Enter test count: 1000
SortBits elapsed: 515
SortGeneric elapsed: 13946
--------------------

Why do we need to "Enter max integer"?
What if max is MaxLongint or MaxUIntValue (I hope you have enough memory to sort five numbers when one of them is close to the end of the range)?
How about negative numbers?

[Pascal]

I was in opposition to including rtl-generics into 3.0.4 (it was considered).

Why? (Due to the workarounds?)

[hnb]

I was in opposition to including rtl-generics into 3.0.4 (it was considered).

Why? (Due to the workarounds?)

Yes, there is also more: FPC has critical bug for class var:

https://bugs.freepascal.org/view.php?id=28911

which is important for usage of case insensitive dictionaries:

https://bugs.freepascal.org/view.php?id=30433

for example:

Code: Pascal  [Select][+][-]

1. StringMap := THashMap<String, TEmptyRecord>.Create(TIStringComparer.Ordinal);

the library should be perfect. There was also small problem with quadratic probing for TOpenAddressingQP<TKey, TValue> (kind of dictionary), which is anyway now fixed. Also was small bug for specific case in dictionaries (related to constref):

https://github.com/maciej-izak/generics.collections/blob/cf4e62eded46a03adce339ddf6552600c3e27d1b/tests/t1.lpr

also fixed (for now only in github version of library). I also need (or Sven) to fix few generics issues. I like to use anonymous methods in final version too. When you think about all of problems then you know that the better choice was not including rtl-generics to 3.0.4.

[bylaardt]

see this:
https://github.com/bylaardt/TOrderList
like a boss style!

[Thaddy]

see this:
https://github.com/bylaardt/TOrderList
like a boss style!

No, like overkill style. There a big difference between an array and a list. Your code is also the slowest for the problem presented (5 items) . KUDOS for parallel, zero for applicability.

[bylaardt]

Hi Thaddy (i miss the          thing and the "grumpy mode on")

I agree! 
too much for 5 items.