AutoIt v3.3.13.8 Beta

[Jon]

File Name: AutoIt v3.3.13.8 Beta

File Submitter: Jon

File Submitted: 22 Jul 2014

File Category: Beta

3.3.13.8 (22nd July, 2014) (Beta)
AutoIt:
- Fixed: Regression in With .Method(...).

- Added: MapRemove()

Click here to download this file

[mLipok]

The problem with HelpFile and "Related" section is still there.

 

Related
(), (), (), (), (), (), (), (), (), (), ()
Example

[guinness]

I fixed the build script. The help files need to be reconstructed that is all.

[mLipok]

Ok.

[TypeIt]

The following code usually crashes.

Local $args = ["CallArgArray", 1, 2]
For $i = 1 To 100
    Call (test, $args)
Next
ConsoleWrite ("didn't crash")

Func test ($arg1, $arg2)
EndFunc

[Melba23]

TypeIt,

After multiple attempts, that code does not crash for me in either 3.3.12.0 or 3.3.13.8. Does it crash for you in earlier versions or just the Beta?

M23

[santeri]

Hello! After version 3.3.13.7 arrays work noticably slower -- from ~11% slower for 1D arrays to ~24% for 3D array...  

And btw, curiously enough, version 3.3.13.6 is slightly faster that 3.3.12.0

version 3.3.12.0: 1138.2299784516

version 3.3.13.6: 1124.6431906223

version 3.3.13.8: 1410.6926015453

Global $t, $d, $dd=0, $a[10000][2][3], $biggest, $smallest
SRandom(12)

;fill with random numbers
$t=TimerInit()
for $i=1 to 100
    for $j=0 to UBound($a)-1
        $a[$j][0][0]=Random(0,10000,1)
    next
next
$d=TimerDiff($t)
ConsoleWrite('filling with random numbers took: '& $d &@cr)

for $i=1 to 10
    $t=TimerInit()
    for $j=1 to 100
        $biggest=$a[0][0][0]
        $smallest=$a[0][0][0]
        for $k=0 to UBound($a)-1
            if $a[$k][0][0]>$biggest then $biggest=$a[$k][0][0]
            if $a[$k][0][0]<$smallest then $smallest=$a[$k][0][0]
        next
    next
    $d=TimerDiff($t)

    $dd+=$d
    ConsoleWrite($i&'\10: '&$d &@cr)
next
ConsoleWrite('----' &@cr& $dd/10 &@cr& '----' &@cr)
exit

[Jon]

TypeIt,

After multiple attempts, that code does not crash for me in either 3.3.12.0 or 3.3.13.8. Does it crash for you in earlier versions or just the Beta?

M23

It crashes in debug mode, so there's something up.

[TypeIt]

Does it crash for you in earlier versions or just the Beta?

Just this and the last beta.

 

After multiple attempts, that code does not crash for me in either 3.3.12.0 or 3.3.13.8.

The following line crashes sometimes. When repeating that statement one hundred times, it usually crashes.

Call (test, $args)

It doesn't crash when the called function expects less than two arguments or too few arguments are passed.

[jchd]

Jon,

Something else for you. DllStructs don't support zero-length struct elements:

Local $t = DllStructCreate("byte[0]")
ConsoleWrite("error = " & @error & "  struct size = " & DllStructGetSize($t) & @LF)
Local $read = DllStructGetData($t, 1)
ConsoleWrite("error = " & @error & "  value read is " & VarGetType($read) & " = " & $read & @LF)
Local $read = DllStructGetData($t, Default)
ConsoleWrite("error = " & @error & "  value read is " & VarGetType($read) & " = " & $read & @LF)

The issue arose when a user tried to crypt/decrypt a zero-length RC4 ciphertext, which is absolutely legal. >The thread is there.

Now what do we do: either introduce a kludge in Crypt.au3 or rely on the possibility that you allow support for zero-length items in dllstructs?

monoceres and I at least are in favor of the later.

[czardas]

 

Hello! After version 3.3.13.7 arrays work noticably slower -- from ~11% slower for 1D arrays to ~24% for 3D array...  

And btw, curiously enough, version 3.3.13.6 is slightly faster that 3.3.12.0

version 3.3.12.0: 1138.2299784516

version 3.3.13.6: 1124.6431906223

version 3.3.13.8: 1410.6926015453

Global $t, $d, $dd=0, $a[10000][2][3], $biggest, $smallest
SRandom(12)

;fill with random numbers
$t=TimerInit()
for $i=1 to 100
    for $j=0 to UBound($a)-1
        $a[$j][0][0]=Random(0,10000,1)
    next
next
$d=TimerDiff($t)
ConsoleWrite('filling with random numbers took: '& $d &@cr)

for $i=1 to 10
    $t=TimerInit()
    for $j=1 to 100
        $biggest=$a[0][0][0]
        $smallest=$a[0][0][0]
        for $k=0 to UBound($a)-1
            if $a[$k][0][0]>$biggest then $biggest=$a[$k][0][0]
            if $a[$k][0][0]<$smallest then $smallest=$a[$k][0][0]
        next
    next
    $d=TimerDiff($t)

    $dd+=$d
    ConsoleWrite($i&'\10: '&$d &@cr)
next
ConsoleWrite('----' &@cr& $dd/10 &@cr& '----' &@cr)
exit

;

Your test code contains unrelated functions, conditionals, assignments and random elements; which don't necessarily give a very clear picture. In any case, after reading about the changes, this is a major concern. If it's 24% slower with 3D arrays, what will it be with more dimensions? Can anyone else confirm these findings?

Edited July 24, 2014 by czardas

[guinness]

I think it's a little unfair to start comparing results like that with a beta version vs stable. I am ignoring the results and concentrating on what is actually important.

[Jos]

I think it's a little unfair to start comparing results like that with a beta version vs stable. I am ignoring the results and concentrating on what is actually important.

Maybe a little harsh response?

It is an observation brought forward and think it is good that these kind of things are voiced, so it at least can be considered/checked to see if this is a logical consequence of the changes or maybe an issue.

Jos

[jchd]

Yeah, on the contrary it can be helpful to draw Jon's attention after his massive array support rewrite.

[Jon]

 

The following code usually crashes.

Local $args = ["CallArgArray", 1, 2]
For $i = 1 To 100
    Call (test, $args)
Next
ConsoleWrite ("didn't crash")

Func test ($arg1, $arg2)
EndFunc

 

Fixed in next beta.  It should never have worked in previous versions either. 

[Jon]

Jon,

Something else for you. DllStructs don't support zero-length struct elements:

Local $t = DllStructCreate("byte[0]")
ConsoleWrite("error = " & @error & "  struct size = " & DllStructGetSize($t) & @LF)
Local $read = DllStructGetData($t, 1)
ConsoleWrite("error = " & @error & "  value read is " & VarGetType($read) & " = " & $read & @LF)
Local $read = DllStructGetData($t, Default)
ConsoleWrite("error = " & @error & "  value read is " & VarGetType($read) & " = " & $read & @LF)

The issue arose when a user tried to crypt/decrypt a zero-length RC4 ciphertext, which is absolutely legal. .

Now what do we do: either introduce a kludge in Crypt.au3 or rely on the possibility that you allow support for zero-length items in dllstructs?

monoceres and I at least are in favor of the later.

That's an invalid struct in C. If the API takes a string an empty string would be byte[1] with a null element or a null ptr depending on the API. Edited May 9, 2015 by Jon

[guinness]

That's an invalid struct in C. If the API takes a string an empty string would be byte[1] with a null element or a null ptr depending on the API.

Noted and fixed in the crypt functions.

Edited July 24, 2014 by guinness

[jchd]

That means we have to review code where a zero-size variable e.g. Binary('') may be passed or result in a buffer of size zero then adjust things with new error codes. Just WinAPI[Ex] is a lot!

[czardas]

I wish I had time to get more involved but I have all cylinders firing at once - a consequence of being self-employed.

[BrewManNH]

Hello! After version 3.3.13.7 arrays work noticably slower -- from ~11% slower for 1D arrays to ~24% for 3D array...  

And btw, curiously enough, version 3.3.13.6 is slightly faster that 3.3.12.0

 

version 3.3.12.0: 1138.2299784516

version 3.3.13.6: 1124.6431906223

version 3.3.13.8: 1410.6926015453

These are the results I got on 4 different versions, one being an older release version.

version 3.3.8.1: 2103.76097548142 < release version

version 3.3.12.0: 1776.77396205806 < release version

version 3.3.13.2: 1769.10959648772 < older beta

version 3.3.13.8: 2160.37051253727 < latest beta