Latest Beta

[GEOSoft]

Can we hold up for a day Jon?

Something has definitly broken and I want to backtrack to see where it happened. I'm not sure if it's in AutoIt or in Scite nor do I know in what beta version it happened.

Everything has been fine with test scripts and any other small scripts I tried but when I got to testing the PCRE Toolkit I ran into a major issue. Not only will it not Run under Scite it won't compile either. Since there are no script breaking changes listed I will have to start going back to previous versions of the beta and see if I can find out when it happened and then try to see what is happening. It will be 3 or 4 hours before I can even start on that part.

[ProgAndy]

Yes, bug in the testscript.

@jchd: Thanks for the script, now my nested DLLStructs are corrected, too.

[GEOSoft]

Okay, I took a few minutes away from service to check this.

The Beta isn't what is directly causing my problem. It is however causing a problem in one file that is #Included in my code. I still don't know why it tried to compile the Beta version and just hung up in the middle and went no place. I had to force the compile (through Scite) to close with End Process.

Also Just an FYI; I have to look at why my processor usage went up to about 90% on 2 of my cores after I forced the process to close but that is probably not AutoIt related. It's just never happened with anything before now.

@jchd

The problem file is one of yours. Get hold of me on IM when you get a chance and I'll give you the details. I probably won't be back here for a while so better with the IM.

[Ascend4nt]

3.3.7.3 crashes on parsing a (meaningless) align directive at the end of a _struct:

_{[autoit]Local $s5 = DllStructCreate("int; STRUCT; short; char; align 1; ENDSTRUCT; char")}

_{Interesting, I just found align on its own or at the end of the string causes crashes on the production v3.3.6.1 as well as the latest betas. The following strings crash AutoIt (adding ';' after align doesn't help):}

"align 4"
"align 4;dword;byte;align 2"

[Jon]

Yeah, fixed in my working copy

[Richard Robertson]

Can anyone out there give me a good reason for structures to change alignment part of the way through?

While on that topic, why are structures padded anyway?

[Jon]

Can anyone out there give me a good reason for structures to change alignment part of the way through?

None really. 99.9999% of the winapi structures are at default alignment. But all these weird tests at least give me the confidence that our padding logic is correct.

When a modern computer reads from or writes to a memory address, it will do this in word sized chunks (e.g. 4 byte chunks on a 32-bit system). Data alignment means putting the data at a memory offset equal to some multiple of the word size, which increases the system's performance due to the way the CPU handles memory. To align the data, it may be necessary to insert some meaningless bytes between the end of the last data structure and the start of the next, which is data structure padding.

[Valik]

Jon, remember to use A_01_configure_build_settings to update the versions. It looks like Au3Check had a manual change. Also for some reason the change to 3.3.7.4 didn't happen for all the files. Not sure if this was an issue with the script, with you not using the script or some other working copy weirdness that prevented the changes from getting committed. Previous version changes appear to have gone okay so I assume you are using the script at least some of the time.

[Jon]

Yeah, I always use the script.

[Valik]

Yeah, I always use the script.

Strange. Something definitely went wrong with the 3.3.7.4 change as you can see from one of my more recent commits. We can chalk the Au3Check one up to probably not having the right year set. Hopefully it was a one-time thing and not a bug in the script.

[trancexx]

Is it possible to change the syntax for the sake of forward compatibility to something like

$sStruct = 'align 2;int;{double};byte'
$stStr = DllStructCreate($sStruct)

$sStruct = 'int;{ptr;int};byte'
$stStr = DllStructCreate($sStruct)

I can understand the need to have the simplest transition as possible with UDFs, but still. And it looks weird to me.

Jokes aside, really.

[jchd]

@Richard,

Can anyone out there give me a good reason for structures to change alignment part of the way through?

Because existing structs already in use in the wild may require it, either in files or memory-mapped I/O where cheating is out of question.

I agree that you can just go with align 1 and insert padding with filler bytes yourself to model any existing struct, but doing so is not only very cumbersome and also prone to potentially serious errors even if you are cautious.

I had once to create thousand of AutoIt structs with sometimes weird layout and I'd have liked to avoid inserting dummies everywhere.

why are structures padded anyway?

Struct arrays and nested structs often require padding, and sometimes with different packing sizes for inner struct(s).

[Jon]

3.3.7.4 (28th May, 2011) (Beta)

- Fixed: More DllStruct alignment fixes and crash.

- Fixed: All VC10 Code Analysis warnings fixed. Check for subtle bugs.

[GEOSoft]

Just a general question about the Betas

Is there any particular reason to have the Scite folder included in the Beta SFX files? Or in the Beta installer for that matter since it doesn't install without the release version if I remember correctly.

As far as I know it will still use the version of Scite installed in the main AutoIt3 folder so having the extra one seems somewhat redundant.

This only came to mind when I thought of sending a PM to Jos about something Scite related that I noticed while I was testing a script.

[Jos]

Just a general question about the Betas

Is there any particular reason to have the Scite folder included in the Beta SFX files? Or in the Beta installer for that matter since it doesn't install without the release version if I remember correctly.

As far as I know it will still use the version of Scite installed in the main AutoIt3 folder so having the extra one seems somewhat redundant.

This only came to mind when I thought of sending a PM to Jos about something Scite related that I noticed while I was testing a script.

This is only valid when you either have the Full version of SciTE4AutoIt3 installed or no changes were made to the lite version supplied with the production version of Autoit3.

Jos

[Jon]

We don't have separate zip commands for the beta/release so scite/lite just gets included by accident. It's not included in the installer though.

[Jon]

Jon, remember to use A_01_configure_build_settings to update the versions. It looks like Au3Check had a manual change. Also for some reason the change to 3.3.7.4 didn't happen for all the files. Not sure if this was an issue with the script, with you not using the script or some other working copy weirdness that prevented the changes from getting committed. Previous version changes appear to have gone okay so I assume you are using the script at least some of the time.

This just happened again. I'd accidentally used the x64 version of autoit to run it and it only half works. Arse.

Edit: Specifically, it's not modifiying the .rc files.

Edit2: Some function in "VersionSetByPath" must be failing as input and output are always identical. This will be fun to track down >.<

Edit3: It works in the Prod version, that makes it a bit easier to find I guess.

Edit4: Need to go out. Found that it occurs between revision 6011-6026

[Valik]

Oooh, I've never tested any of those scripts as x64. At least we have a clue now, though.

[Jon]

Need to go out. Found that it occurs between revision 6011-6026

[Valik]

Oh, so it's a bug in AutoIt itself in one of the betas that only happens with x64? In other words, AutoIt is broke, not the script.