AutoIt in a different perspective

[czardas]

Ternary is too. This reminds me, I recently discovered some mistakes in the (my own) documentation, I need to update it. It's a job for later on today.

Edited September 18, 2012 by czardas

[guinness]

I recently discovered some mistakes in the documentation, I need to update it. It's a job for later on this today.

Great. Just post in the help file thread I would look at it in due course.

Edit: Whoops! I re-read it and realised it was your documentation.

Edited September 18, 2012 by guinness

[skin27]

High order functions, objects, threads etc are great when writing larger programs with AutoIt. In practice most of the time I just needed a simple code snippet (which is mostly available on the wiki if not in the help file).

Another perspective is not from a language construct, but from a first time programmer (where nothing is implied). For example I had some moments where I was confused on its Basic-like syntax. Some examples:

1) Equality

When you compare something it must be of the same type, for example

; If int = int then true

If 1 = 1 then ConsoleWrite("true")

; Ok, that works as expected. However the following valid statement returns always false

If 1 = not(2) then ConsoleWrite("true")

; So I compared an int with the boolean false, so I have to write something like this:

If 1 <> 2 then ConsoleWrite("true")

or

If StringCompare ( "1", "2", 1) <> 0 then ConsoleWrite("true")

To get equality between int, string. Such syntax would be clearer to me

If 1 not = 2 then ConsoleWrite("true")

or

If 1 != 2 then ConsoleWrite("true")

2) arrays

To work with arrays you can typically use Basic statements like Ubound, Dim, Redim

With the current use function naming conventions I would opt for

Ubound - 1 --> ArraySize

ReDim --> ArrayResize

Which new programmer would search for Ubound or ReDim? I also got confused that most of the time you can use array[0] to get the array size, while on other occasions you need Ubound.

Side note: I would like to work with (i.e. Vector) collections, instead of just arrays. Needs less low level work.

3) The following seems to me some historic usage of the Basic syntax.

For example the conditional statement have the following syntax

If --> EndIf

Select --> EndSelect

Switch --> EndSwitch

With --> EndWith

But loops

While --> WEnd

For --> Next

Do --> Until

When just looking at logic of the loop this seems logical, but when looking to the whole language why not do the following:

While --> EndWhile

For...To --> EndFor

Foreach ... in .. EndForeach

Do --> EndDo

If the logic on loops needs to be kept then I would suggest to use While --> Done, instead of While --> WEnd

3) Besides includes it would be nice to define a package/project, so when working with multiple file the init (main) file will always be started.

4) Template expression

This would make codes with multiple lines easier to work with. Like I posted on example script: ression

New JVM languages like Kotlin, Xtend and Ceylon have good support for this.

5) RunDos

When I first started to use _RunDos I would expect that it equals to Run(@ComSpec & " /c " & 'commandName', "", @SW_HIDE). Then there is ConsoleWrite/ConsoleRead which only works in a CUI. Found it all a bit confusion when working with the console. Couldn't this be reworked (like Mat's UDF for this)?

Ok, in my examples I still wrote about some language constructs like packages, collections and template expressions, but I hope the examples on a different perspective from a newbie point-of-view adds also something to this discussion.

[jvanegmond]

3) 'End' keyword to close all control structures. Preprocessor does an automatic conversion to correct keyword to match last used control structure. For example, input:

While True
Sleep(500)
End

Preprocesses into:

While True
Sleep(500)
WEnd

It's a trivial thing, but can be convenient.

4) Multi-line strings. Why didn't I think of that? Pick any syntax (up for debate on this), I would prefer:

$a = [[This is
a multi
line string.]]

Preprocesses into:

$a = "This is" & @CRLF & _
"a multi" & @CRLF & _
"line string."

@CRLF, @LF or @CR can be determined by what characters you use in your source file.

[dany]

4) Multi-line strings. Why didn't I think of that? Pick any syntax (up for debate on this), I would prefer:

$a = [[This is
a multi
line string.]]

Preprocesses into:

$a = "This is" & @CRLF & _
"a multi" & @CRLF & _
"line string."

Like HereDoc strings. Yea that would be a very nice addition, got my vote. As far as syntax goes, [[Lua or the more unofficial standard <<HEREDOC style, well, I'd go for the <<HEREDOC style, Lua is somewhat ambiguous imho as it looks like an array assignment.

$a = <<HEREDOC_END
This is
a multi
line string.
HEREDOC_END

[James]

I +1 Heredocs. Since variable identifier $ is now optional, it should be mandatory to wrap variables to be printed in mustaches {{}} or even how Ruby does it #{variable}

[Mat]

People have done this before, to implement features such as classes using a pre-processor.

[jvanegmond]

People have done this before, to implement features such as classes using a pre-processor.

[guinness]

The preprocessor converts out constant values.

; Before
Local Const $iValue = 10
MsgBox(4096, '', $iValue)

; After
MsgBox(4096, '', 10) ; Replaces $iValue with the respective value.

[jvanegmond]

The preprocessor converts out constant values.

; Before
Local Const $iValue = 10
MsgBox(4096, '', $iValue)

; After
MsgBox(4096, '', 10) ; Replaces $iValue with the respective value.

Doesn't AutoIt do this internally in the lexing or parsing stage?

[guinness]

Doesn't AutoIt do this internally in the lexing or parsing stage?

It's something I have wondered in the past.

[jvanegmond]

It's something I have wondered in the past.

Poking around in v3.3.8.1 says no.

Edit: I'm not a fan of preprocessing for speed, though. It's extremely hard because it's hard to predict. It also puts focus in the wrong place. It distorts responsibilities.

Edited September 19, 2012 by Manadar

[Mat]

Poking around in v3.3.8.1 says no.

Eval("iValue") probably says no too.

[trancexx]

Const var is just like non-const one. The only difference is different flag which is set and read by the parser. Lexer makes no difference.

[czardas]

@CRLF, @LF or @CR can be determined by what characters you use in your source file.

Perhaps I misunderstand this. You don't want to lose control of formatting: considering there are inconsistancies between third party applications which could cause no end of problems. Thinking about the code boxes on this forum, as an example.

Edited September 20, 2012 by czardas

[guinness]

Const var is just like non-const one. The only difference is different flag which is set and read by the parser. Lexer makes no difference.

Thanks for clarifying. It's what I thought happened.