New SciTE4AutoIt3 available with SciTE v1.75

[Jos]

Scite4autoit3 installer install log. Can't find it.

Is is saved somewhere?,

Or do I need to add a startup option to get(keep) it?

On the side observation:

- after (re)installing scite4autoit3 the [scite config].[loaded autoit3 definitions] setting switches from [beta v3.2.11.7] to [Production v3.2.10,0]

Intensional ?

Thanks

Don't think NSIS has an standard option for a log file.

EDIT: Found a DUMPLOG plugin which will now dump the info from the window into the scite program directory and is called install.log.

I will have a look at the Defs update issue in the updatedefs.au3 script which runs at the end of the SciTE4AutoIt3 installer and fix that for the enxt release.

Jos

Edited April 19, 2008 by Jos

[Jos]

I have noticed that when i had installed a new scite4autoit, the new "Autoiscript" in the new menu when you right click in the desktop dissapears.

The Right-Mouse-click/New/AutoIt3Script is there for me and doesn't dissapear when running the SciTE4AutoIt3 installer.

Jos

[wraithdu]

Not sure if it's already on your dev roadmap, but the same path search changes that you made to Obfuscator and SciteConfig need to also be made to AutoIt3Wrapper/Gui. I noticed this in your new beta versions. BTW, the GUI beta source is for 1.10.1.8, but the binary is 1.10.1.7.

[Jos]

Not sure if it's already on your dev roadmap,

My roadmap? I know what you mean but it does sound a bit "too official" for this project.

, but the same path search changes that you made to Obfuscator and SciteConfig need to also be made to AutoIt3Wrapper/Gui. I noticed this in your new beta versions. BTW, the GUI beta source is for 1.10.1.8, but the binary is 1.10.1.7.

Now to your question: What are you exactly thinking of that requires the same search path changes in AutoIt3Wrapper and the GUI?

Currently AutoIt3Wrapper assumes to be located in a subdirectory of SciTE which in turn is expected to be a subdir of the autoi3 program directory.

"C:\program files\AutoIt3\SciTE\AutoIt3wrapper". AUtoIt3Wrapper will strip "\SciTE*" leaving the AutoIt3 program directory.

Thanks for the Info on the GUI. There are no changes between these versions but its just to keep the gui in sync with the AutoIt3Wrapper program.

Jos

[wraithdu]

The directory problems only surface if running the GUI with Scite's Compile command, ie without any commandline info (so the GUI has to determine paths on its own).

1. The SciTE directory is read from the registry only, so it is not updated correctly if running without a proper install (portably), Lines 61-62. This causes missing version info for Tidy (Line 930), Obfuscator (Line 960), and a couple of cvswrapper problems, as the file won't be found (Lines 985, 988).

2. While the AutoItDir is found as you mentioned (Lines 254-264), the value of $CurrentAutoIt_InstallDir comes only from the commandline or the registry (Lines 65-66 or Lines 144-145). So paths to Au2exe (Line 727) and Au3Check (Line 900) are not correct (the version info in the GUI is missing, for example).

I think that's all. You can fix #2 by setting the value of $CurrentAutoIt_InstallDir from $INP_AutoitDir after the checks in Lines 254-264.

#1 will require the full search (or something thereabout) for the correct location of SciTE.

[Jos]

You are right, support for Const wasn't build in yet but is now. Give the version in the Beta directory a try.

Dank,

Jos

Edited May 5, 2008 by Jos

[Jos]

mmm... You must be the only one using the script reporting option because continuation lines never worked for any of the declare functions (Dim/Glocal/Local/Const).

Fixed that too.

Thanks

[James]

mmm... You must be the only one using the script reporting option because continuation lines never worked for any of the declare functions (Dim/Glocal/Local/Const).

Fixed that too.

Thanks

Don't say it like a bad thing

How do you turn this "script reporting" option on?

[Jos]

Don't say it like a bad thing

How do you turn this "script reporting" option on?

Let me see how I say this politely: Did you RTFM ?

- or -

Ever done Ctrl+F7? The GUI will show you the available Tidy options supported.

Sometime I wonder why I ever bother writing a helpfile. Its the worsed part of creating an application while knowing most wont read it anyways.

Edited May 7, 2008 by Jos

[Zedna]

Sometime I wonder why I ever bother writing a helpfile. Its the worsed part of creating an application while knowing most wont read it anyways.

Your helpfile isn't useless. I use it quite often to check exact AutoIt3Wrapper syntax/options

[Jos]

Was just kidding folks... wanted to make a point to a regular whom I am sure is deeply ashamed and doesn't dare to reply anymore.

(just kidding James)

Edited May 8, 2008 by Jos

[Jos]

SciteConfig start-menu link point to 'SciTE\SciteConfig.exe' instead of 'SciTE\SciTEConfig\SciteConfig.exe'

curtesy of ccleaner.

Fix in the NSIS installer for the next version.

[Jos]

Hi, me again.

More a personal request than a issue.

Could the scite4autoit tool function "Trace: Add Func Trace Lines" be made to ignore commented-out function lines.?

Currently it adds them to any line that contains the text "Func name(*)".

M.v.G

This is one written by Valik.

I thought this one was brought up before but need to check that.

Jos

[Valik]

Jos, the same issue was brought up for something else which I don't remember. I'll take a look and fix it a bit later.

[Zedna]

BUG:

Obfuscator /striponly removes OnAutoItExit() function from script:

Here is testing script:

When you comment first two lines messagebox appears when you run compiled EXE

#AutoIt3Wrapper_run_obfuscator=y
#Obfuscator_parameters=/so

Func OnAutoItExit()
  msgbox(0,'',@ScriptName)
EndFunc

In this case script_obfuscated.au3 is empty.

Edited May 13, 2008 by Zedna

[Valik]

Hi, me again.

More a personal request than a issue.

Could the scite4autoit tool function "Trace: Add Func Trace Lines" be made to ignore commented-out function lines.?

Currently it adds them to any line that contains the text "Func name(*)".

M.v.G

It does ignore them, at least in my example.

I find I say this too much, but what version and where's the script to reproduce? You've neglected to provide that information and as usual the issue can't be reproduced. The following script works as expected, no trace lines are inserted:

#cs
Func Commented Out
EndFunc
#ce

[Jos]

It does ignore them, at least in my example.

I find I say this too much, but what version and where's the script to reproduce? You've neglected to provide that information and as usual the issue can't be reproduced. The following script works as expected, no trace lines are inserted:

#cs
Func Commented Out
EndFunc
#ce

Give this one a try. your example doesn't contain proper Func lines since they lack the brackets:

#cs
    Func Test()
    EndFunc
#ce
;
;Func Test()
;EndFunc
;
;~ Func Test()
;~ EndFunc

[Valik]

I clearly wasn't thinking this morning.

[Zedna]

Now when testing new beta Obduscator I noticed strange behaviour of AutoIt3Wrapper:

When I add icon esulted EXE size is about 25KB smaller then without this icon!

Icon is standard ICO file 766 bytes in size.

Here is testing script:

#AutoIt3Wrapper_icon=my_icon.ico

#AutoIt3Wrapper_run_obfuscator=y
#Obfuscator_parameters=/so

Func OnAutoItStart()
  msgbox(0,'1',@ScriptName)
EndFunc

Func OnAutoItExit()
  msgbox(0,'2',@ScriptName)
EndFunc

Just try to compile it with/without

#AutoIt3Wrapper_icon=my_icon.ico

EXE size on AutoIt 3.2.10:

without icon:252 KB

with icon:227 KB

EDIT:

Nevermind! I discovered reason with help of reshacker.

My simple 766 byte icon replaced default icon group with standard autoit icons in various sizes/depths

There is 9 icons with 25 KB summary size

EDIT2:

So if you want to shrink output EXE size by 25 KB just use your own simple icon!

Edited May 13, 2008 by Zedna

[Valik]

Jos, this is not a trivial fix. Looking back, it was AutoItGotoDefinition which got the "don't go to functions in comments treatment". Attempting to use the same logic does not work with the Trace functionality. The reason's a rather evil "gotcha". I ensure there are no existing function trace statements before I add new statements. This causes SciTE to forget it's styling information. By the time I get the position of the match, SciTE is just reporting 0 for all styles.

This bug exposes two problems. First and most obviously, the function Common:ReplaceDocByPattern() is significantly flawed when used in this manner. It causes styling information to be reset which may not be desired. The second issue is performance, adding trace lines actually calls into Common:ReplaceDocByPattern() 3 times in total which means the entire document is reset 3 times. On a large file, I imagine this is performance hell.

The solution isn't trivial, as I mentioned. I'm thinking about writing a class to encapsulate the same functionality as Common:ReplaceDocByPattern() except it's going to require a few more function calls but it's also going to be transaction based. The general flow would be something like:

DocTransaction:Begin() -- Loads the document into a variable
 DocTransaction:ReplaceByPattern() -- Replaces the text only in the variable, the document isn't changed yet.
 CallFunctionWhichAlsoUsesReplaceByPattern()
 DocTransaction:Finalize() -- Update the document

Calling Begin() and Finalize() is reference counted so the document is only updated when the count hits 0. Means other functions can be called between Begin() and Finalize() which also use those functions but a document update won't happen. The only other gotcha I see is there needs to be some sort of TranslatePos() function which takes a pos in the modified document and translates it to a position in the original document so that things like editor.StyleAt() will work. This part sounds most tricky. May be do-able, though.

If I go that route, I will, of course, also provide a single function which encapsulates the entire sequence into a single function call for convenience, although it will, of course still be just as "broken" if used the same way as what we have now.