Au3Check: parentheses and square brackets unbalanced - and wrong number of arguments

[c.haslam]

I (perhaps I am not alone) have a bad habit of forgetting to key a closing parenthesis to match an opening parenthesis. A fair bit of the time, when I do, Au3Check tells me about errors that are not errors. I am then left to discover what my error really is. It is always reported, but is often not the first line with a red box under it. Perhaps others have this experience.

So I suggest that one of the first things Au3Check should do is to check for balanced parentheses on each line of code. And, perhaps, if there are unbalanced parentheses, Au3Check should only report them. I make the same suggestion for square brackets.

I hope that giving unbalanced priority would not be a huge change in Au3Check.

While I am on the subject of Au3Check, if my function foo() takes 4 parameters, and I call it once with 4 arguments and, further down in my code, I call it with 3, Au3Check catches my error, but tells me the line number where foo() is called correctly, rather than the line number where I called foo() with only 3 arguments. It would be nice if Au3Check pointed me to the line where I made an error.

 

Thoughts?

 

Edited June 27, 2015 by c.haslam

[mLipok]

Show example code.

[jchd]

About the mismatched number of arguments, that would require Au3Check to perform like a compiler, while it works like an interpretor. Only when the declaration is reached, Au3Check can verify if previous function calls were valid or not.

If yopu insist on getting the error message right and pointing to the first misuse, then place all your functions before their invokations (refactor your script à la Pascal).

[c.haslam]

The problem is that Au4Check points to a valid function call, not an invalid one.

[jchd]

c(1, 2)
c()

Func c()
    Return 0
EndFunc

c(1)

a()
a(1)
a(1, 2)
a(1, 2, 3)

b()
b(1)
b(1, 2)

Func a($p1 = 0, $p2 = 0)
    Return($p1 + $p2)
EndFunc

Func b($p1 = 0)
    Return($p1 + 1)
EndFunc

Which failure do you see here?

[trancexx]

The OP talks about this jchd:

a()
a(1, 2)

Func a()
EndFunc

 

[c.haslam]

jchd

Yes, Au3Check does its work properly on your script, by reporting an error on line 1. But if I reverse lines 1 and 2, Au3Check points me to line 1 (now c()). It should point to line 2 (now c(1, 2)).

This is not a problem with a short script, but it is annoying when the script is long, and the calls to c() may be hundreds of lines apart. One has to use Ctrl_F and F3 to find which call is bad.

PS I just edited my post to correct unbalanced parentheses!

Edited June 28, 2015 by c.haslam

[jchd]

I agree and the new editor played we a bad joke by ripping out a sentence in my previous answer, where I wrote that the first invocation of a function (when the func declaration isn't previously encountered) dictates where the error is pointed out.

I agree that this causes some inconvenience but given how Au3Check currently works (line after line), it isn't surprising. Working out precisely the wrong call(s) to a function would require Au3Check to make a note for each invocation of yet undefined functions with its number of arguments to finally, when encountering the actual function definition, be able to point you to wrong calls. In large scripts that would certainly represent a large amount of data. It's certainly possible, but this is AFAICT not the way this utility works, would require a serious rewrite and I'm not sure Jos is willing to undergo such task.

Again, if you want to guard yourself against misleading error pointers, refactor your scripts in the Pascal style, with every function declaration preceding it's invocation(s).

[mLipok]

Sorry to intrude in this interesting discussion.

 

.... refactor your scripts in the Pascal style, with every function declaration preceding it's invocation(s).

@jchd

Currently I had the idea that all "functions" in each of my script were as UDF files , the main file would contain only #include region and after this Function call inside the main loop. Is this idea is the realization of your statements ?

btw.
Is such a concept (my concept idea) is correct from the point of view of code readability  EDIT: and yours experiences. ?

mLipok

Edited June 27, 2015 by mLipok

[jchd]

Smells like Pascal and quiche eaters

[c.haslam]

Thank you all for your thoughts (and time!)