Assignment operators ???

[guinness]

For a long time I was under the impression $sString= "100" and using Int() was "casting", but it's actually parsing. Now $iInt and using Number() would be considered casting.

[Ascend4nt]

Does this seem as counter intuitive to anyone else or is it just me?

 

Of course its counterintuitive . Logically zero or any number doesn't equal a string that isn't itself convertible to a number.  Anyone that tells you differently just 'accepts' the way a programming language works and doesn't have the ability to look at the situation objectively.

In a typed language, comparing a string to a integer would cause a fatal error. You should think yourself lucky that AutoIt does not do the same.

I don't know what typed languages you use, but most any proper language will fail to compile or report an issue while trying to parse it.

 

This "string" = 0 nonsense may make sense to people that just "accept" the language quirks, but I see it as a logic failure.  Perhaps it wasn't well thought out, but at the very least the programmers could have taken inspiration from something like Javascript, which will produce a NaN (not-a-number) result for strings which can't be converted to numbers. Otherwise it's just nonsense.  And not very consistent with another quirk - where boolean tests of "" produce False and "string" produces True, which can further produce the values 0 and 1..

But ah, it's too late to fix the language I suppose. Still, these string comparison quirks of the language should be documented somewhere.

[jaberwacky]

And not very consistent with another quirk - where boolean tests of "" produce False and "string" produces True, which can further produce the values 0 and 1..

 

Ah, yes, I think this is the source of my confusion.

[jchd]

 

Still, these string comparison quirks of the language should be documented somewhere.

They are, under Language Reference - Datatypes

[Ascend4nt]

They are, under Language Reference - Datatypes

I see nothing discussing comparison or equality operators.  If you are referring to this:

If a string is used as a number, an implicit call to Number() function is done. So if it doesn't contain a valid number, it will be assumed to equal 0.

then you are surely missing something.  The above is referring to using a string "as a number", not comparing two variant types. And this is the fundamental problem with the language design - variant types should not need to have a programmer carefully consider something as fundamental as equality statements. Either something is the same logical value or it isn't. And in the case discussed here, it very obviously isn't.  But I digress.. as I said before, the documentation regarding comparisons isn't there, period. *

Also, it seems odd that the quoted text wouldn't be included in the Strings and Numbers section or at least referenced from there.

*edit: see jchd's post below (Operators section, not Datatypes)

Edited June 5, 2014 by Ascend4nt

[czardas]

variant types should not need to have a programmer carefully consider something as fundamental as equality statements.

 

I understood the original intention was to overcome complications encountered with mismatching data types, so that anyone with little, or no, programming experience could start to code without overly concerning themselves with such things. As a consequence this appears to have introduced a number of inconsistencies which may sometimes have the opposite effect - creating more confusion than is perhaps warranted. I would say it's an ambitious and honorable objective, but there also appear to be certain drawbacks. 

Edited June 5, 2014 by czardas

[jchd]

A (short) discussion about comparing mixed types can be found as the note at the bottom of Language Reference - Operators

« Note: Care is needed if comparing mixed datatypes, as unless the == case-sensitive string operator is used, mixed comparisons are usually made numerically. Most strings will be evaluated as 0 and so the result may well not be the one expected. It is recommended to force the items being compared into the same datatype using Number/String before the comparison. »

[Ascend4nt]

jchd, I had to search for that, as I had never seen that before.  Since I hadn't upgraded to the new AutoIt (constants nightmare), it wasn't in my version (3.3.10.2).  However, I see it now - but why is it at the very bottom of the page, and with no references to it in the Comparison Operators section of the table?  I'd reckon people will stumble upon that by accident with no other reason to scroll to the bottom!  But hey, its a start. Putting some reference to it in the Comparsion Operators section of the table, as well as a mention in the Datatypes topic (where you originally directed me) would be good as well.

czardas, mismatching datatypes is exactly what a non-typed language needs to put special focus on.  There's nothing special about AutoIt in this regards - there are plenty of non-typed script languages. If they don't get it right, they wind up with confused programmers complaining about a quite obvious flaw in the language design. I believe PHP suffers the same problem (and that's a language I rather dislike).

[Melba23]

Hi,

As certain people seem to be getting extremely wound up about something that has existed in AutoIt since I started using it many years ago, I have amended the Help file <Operators> page so that the standard "=" operator section points to the "mixed datatypes" note at the bottom of the same page. I hope that will help return any elevated blood pressure readings to a safer level.

M23

[guinness]

"Constants nightmare" is a little harsh don't you think? Anyway, it's not really up for discussion anymore.

Edited June 5, 2014 by guinness

[Ascend4nt]

Melba, good to hear about the Help-file adjustment. I think 'elevated blood pressure' is a misinterpretation on your behalf, or perhaps your own feelings on the matter filtering through. All this is is a relevant discussion on a language flaw (no matter how long the problem has been a part of AutoIt).

[Melba23]

Ascend4nt,

You are beginning to make a habit of "interpreting" my feelings - kindly stick to your interpreting your own from now on please.

M23

[Ascend4nt]

Melba, how can I be in a habit of something I could care less about. If you have 'feelings', don't ruin this thread with them - PM me instead.

[czardas]

"Constants nightmare" is a little harsh don't you think?

 

I think this depends on how you write your code. I don't know where to even begin to untangling the UDF libraries. Not that I'm complaining, but I'm not a fan of dependencies and try to keep them to a minimum. Some of my scripts are both forwards and backwards compatible because I ripped out the functions I needed, to prevent changes in the libraries breaking anything. Not doing this could have far reaching consequences for both me and anyone who might use my code in the future. I also like to fine tune my scripts, so I remove things like error checks etc... This is my personal preference and not a criticism.

Edited June 6, 2014 by czardas

[Melba23]

Ascend4nt,

 

don't ruin this thread with them - PM me instead

Pot to kettle, over.

M23

Edit:

Just in case it passed you by

Edited June 6, 2014 by Melba23