AutoIt v3.3.16.0 - Doubles greater then of Len 10 break SRandom()

[Rurorita]

 

Local $nDoubleNo1 = Number("50853881441621333029", 3)
Local $nDoubleNo2 = Number("93541875452114878455", 3)
Local $nDoubleNo3 = Number("1234567890", 3)

ConsoleWrite("DoubleNo1 (Type:" & VarGetType($nDoubleNo1) & ") is: " & $nDoubleNo1 & @CRLF)
ConsoleWrite("DoubleNo2 (Type:" & VarGetType($nDoubleNo2) & ") is: " & $nDoubleNo2 & @CRLF)
ConsoleWrite("DoubleNo3 (Type:" & VarGetType($nDoubleNo3) & ") is: " & $nDoubleNo3 & @CRLF)

ConsoleWrite(@CRLF)

ConsoleWrite("Feeding DoubleNo1 into SRandom" & @CRLF)
SRandom($nDoubleNo1)

ConsoleWrite("First Random: " & Random(100, 999, 1) & @CRLF)
ConsoleWrite("Second Random: " & Random(100, 999, 1) & @CRLF)

ConsoleWrite(@CRLF)

ConsoleWrite("Feeding DoubleNo2 into SRandom" & @CRLF)
SRandom($nDoubleNo2)

ConsoleWrite("First Random: " & Random(100, 999, 1) & @CRLF)
ConsoleWrite("Second Random: " & Random(100, 999, 1) & @CRLF)

ConsoleWrite(@CRLF)

ConsoleWrite("Feeding DoubleNo3 into SRandom" & @CRLF)
SRandom($nDoubleNo3)

ConsoleWrite("First Random: " & Random(100, 999, 1) & @CRLF)
ConsoleWrite("Second Random: " & Random(100, 999, 1) & @CRLF)

; what

The first two doubles, even so they are entirely different, when feed into SRandom() create the same Random() results, when they actually should create different results. This seems to happen always once the Double is greater then of Len 10. Which is why the third double then returns a different result. So no matter what the number is, once its longer then 9, SRandom() seems to break. So if one would now add one additional number to No3 then the result would also be the same.

DoubleNo1 (Type:Double) is: 5.08538814416213e+19
DoubleNo2 (Type:Double) is: 9.35418754521149e+19
DoubleNo3 (Type:Double) is: 1234567890

Feeding DoubleNo1 into SRandom
First Random: 993
Second Random: 958

Feeding DoubleNo2 into SRandom
First Random: 993
Second Random: 958

Feeding DoubleNo3 into SRandom
First Random: 849
Second Random: 975

A UDF of mine depends on SRandom with bigger nummers, which is why i noticed this.

[jchd]

If you'd had taken the pain to just look at the help on SRandom, you'd have discovered that:

Quote

Seed value for random number generation. Number between -2^31 and 2^31-1

O, surprise!

[Rurorita]

mhm, it did work before

Edited March 7, 2022 by Rurorita

[jchd]

The hint in the help is there: -2^31 and 2^31-1
That's exactly the range of a signed 32-bit integer.

If you follow the clue, you'd use integers instead of doubles, and, ô surprise #2:

Local $nDoubleNo1 = 64654645644
Local $nDoubleNo2 = 84253454868
Local $nDoubleNo3 = 12345678903

ConsoleWrite("DoubleNo1 (Type:" & VarGetType($nDoubleNo1) & ") is: " & $nDoubleNo1 & @CRLF)
ConsoleWrite("DoubleNo2 (Type:" & VarGetType($nDoubleNo2) & ") is: " & $nDoubleNo2 & @CRLF)
ConsoleWrite("DoubleNo3 (Type:" & VarGetType($nDoubleNo3) & ") is: " & $nDoubleNo3 & @CRLF)

ConsoleWrite(@CRLF)

ConsoleWrite("Feeding DoubleNo1 into SRandom" & @CRLF)
SRandom($nDoubleNo1)

ConsoleWrite("First Random: " & Random(100, 999, 1) & @CRLF)
ConsoleWrite("Second Random: " & Random(100, 999, 1) & @CRLF)

ConsoleWrite(@CRLF)

ConsoleWrite("Feeding DoubleNo2 into SRandom" & @CRLF)
SRandom($nDoubleNo2)

ConsoleWrite("First Random: " & Random(100, 999, 1) & @CRLF)
ConsoleWrite("Second Random: " & Random(100, 999, 1) & @CRLF)

ConsoleWrite(@CRLF)

ConsoleWrite("Feeding DoubleNo3 into SRandom" & @CRLF)
SRandom($nDoubleNo3)

ConsoleWrite("First Random: " & Random(100, 999, 1) & @CRLF)
ConsoleWrite("Second Random: " & Random(100, 999, 1) & @CRLF)

 

[Rurorita]

1 minute ago, jchd said:

The hint in the help is there: -2^31 and 2^31-1
That's exactly the range of a signed 32-bit integer.

If you follow the clue, you'd use integers instead of doubles, and, ô surprise #2:

Local $nDoubleNo1 = 64654645644
Local $nDoubleNo2 = 84253454868
Local $nDoubleNo3 = 12345678903

ConsoleWrite("DoubleNo1 (Type:" & VarGetType($nDoubleNo1) & ") is: " & $nDoubleNo1 & @CRLF)
ConsoleWrite("DoubleNo2 (Type:" & VarGetType($nDoubleNo2) & ") is: " & $nDoubleNo2 & @CRLF)
ConsoleWrite("DoubleNo3 (Type:" & VarGetType($nDoubleNo3) & ") is: " & $nDoubleNo3 & @CRLF)

ConsoleWrite(@CRLF)

ConsoleWrite("Feeding DoubleNo1 into SRandom" & @CRLF)
SRandom($nDoubleNo1)

ConsoleWrite("First Random: " & Random(100, 999, 1) & @CRLF)
ConsoleWrite("Second Random: " & Random(100, 999, 1) & @CRLF)

ConsoleWrite(@CRLF)

ConsoleWrite("Feeding DoubleNo2 into SRandom" & @CRLF)
SRandom($nDoubleNo2)

ConsoleWrite("First Random: " & Random(100, 999, 1) & @CRLF)
ConsoleWrite("Second Random: " & Random(100, 999, 1) & @CRLF)

ConsoleWrite(@CRLF)

ConsoleWrite("Feeding DoubleNo3 into SRandom" & @CRLF)
SRandom($nDoubleNo3)

ConsoleWrite("First Random: " & Random(100, 999, 1) & @CRLF)
ConsoleWrite("Second Random: " & Random(100, 999, 1) & @CRLF)

 

As said, it did work in the previous autoit stable

[jchd]

And what do you think is failing?

[Rurorita]

1 minute ago, jchd said:

And what do you think is failing?

I dont know, im just a amateur who has noticed the change

[jchd]

Maybe previous releases were more permissive than the latest (today's) one, but if you violate the specification expect undefined behavior: it may work or not because this isn't part of the contract.

[Rurorita]

Just now, jchd said:

Maybe previous releases were more permissive than the latest (today's) one, but if you violate the specification expect undefined behavior: it may work or not because this isn't part of the contract.

Thanks anyway, i switch to Integers

[jchd]

I confess that it would be beneficial to have 64-bit integers or even doubles supported but for that we currently have to rely on external feature, like the latest Crypt* functions.

[mLipok]

21 minutes ago, jchd said:

Maybe previous releases were more permissive than the latest (today's) one, but if you violate the specification expect undefined behavior: it may work or not because this isn't part of the contract.

You mean it used to work as it did (not quite right) and now (in Au3.3.3.16.0) it has been improved or even fixed?

Something like a decade ago fix for adding Null support?

As a side note:
The addition of Null support hit a few of my projects because ADO started returning Null instead of 0 (or something like this ... I do not remember exactly)
 

Edited March 7, 2022 by mLipok

[jchd]

It seems it has been just made compliant with the specification, which it wasn't quite in previous releases.

[mLipok]

4 minutes ago, jchd said:

It seems it has been just made compliant with the specification, which it wasn't quite in previous releases.

Is it this:

https://www.autoitscript.com/trac/autoit/ticket/3760

?

Edited March 7, 2022 by mLipok

[jchd]

Maybe. I remember having posted at least two trac tickets about large ints but I don't know what change in code their fix did cause, nor their consequences.

[Rurorita]

4 hours ago, jchd said:

I confess that it would be beneficial to have 64-bit integers or even doubles supported but for that we currently have to rely on external feature, like the latest Crypt* functions.

So doubles no longer work like before, but Int64 do now. I think this was also different before

Local $nNo1 = 6465464564422266666
Local $nNo2 = 9223372036854775807 ; max
Local $nNo3 = 4565615152178932588

ConsoleWrite("No1 (Type:" & VarGetType($nNo1) & ") is: " & $nNo1 & @CRLF)
ConsoleWrite("No2 (Type:" & VarGetType($nNo2) & ") is: " & $nNo2 & @CRLF)
ConsoleWrite("No3 (Type:" & VarGetType($nNo3) & ") is: " & $nNo3 & @CRLF)

ConsoleWrite(@CRLF)

ConsoleWrite("Feeding No1 into SRandom" & @CRLF)
SRandom($nNo1)

ConsoleWrite("First Random: " & Random(100, 999, 1) & @CRLF)
ConsoleWrite("Second Random: " & Random(100, 999, 1) & @CRLF)

ConsoleWrite(@CRLF)

ConsoleWrite("Feeding No2 into SRandom" & @CRLF)
SRandom($nNo2)

ConsoleWrite("First Random: " & Random(100, 999, 1) & @CRLF)
ConsoleWrite("Second Random: " & Random(100, 999, 1) & @CRLF)

ConsoleWrite(@CRLF)

ConsoleWrite("Feeding No3 into SRandom" & @CRLF)
SRandom($nNo3)

ConsoleWrite("First Random: " & Random(100, 999, 1) & @CRLF)
ConsoleWrite("Second Random: " & Random(100, 999, 1) & @CRLF)

 

No1 (Type:Int64) is: 6465464564422266666
No2 (Type:Int64) is: 9223372036854775807
No3 (Type:Int64) is: 4565615152178932588

Feeding No1 into SRandom
First Random: 342
Second Random: 407

Feeding No2 into SRandom
First Random: 185
Second Random: 889

Feeding No3 into SRandom
First Random: 844
Second Random: 833

 

[argumentum]

<rant>

Legacy !.
There are thing that, to me, are more practical as UDF than internal. 
_Crypt_GenRandom() works and if not happy with it it can be changed by the community unlike the internal function that takes issuing a new version of AutoIt. Same with @OSVersion, if it was a UDF it'd be easier/faster to attend to than depending on the internal one that again, requires a recompile of AutoIt and a new version that may otherwise not be needed, if just for that internal function.

</rant>

[Jos]

Nobody is stopping you/others from adding useful UDFs to the standard library. Happened before that a udf was added, similar to an internal function, with extended functionality. 😉 

[jessicaagu]

You mean it used to operate as it did (not quite properly) and now it has been enhanced or even corrected (in Au3.3.3.16.0)?

Is there a remedy for adding Null support from a decade ago?

ADO started returning Null instead of 0 when Null support was added, which impacted a handful of my projects (or something like this ... I do not remember exactly)

[mLipok]

3 hours ago, jessicaagu said:

Is there a remedy for adding Null support from a decade ago?

ADO started returning Null instead of 0 when Null support was added, which impacted a handful of my projects (or something like this ... I do not remember exactly)

Why do you want to broke things that was fixed?

Edited April 16, 2022 by mLipok

[jpm]

Back to your first pb.

You cannot expect that  the your number which is converted to an integer be not between ( 2^63 2^63 -1) as SRandom need to have an integer as argument

if you use $nDoubleNo1 = 9223372036854775807 and $nDoubleNo2 = 9223372036854775806 everything is OK