Spaces are allowed here ?

[Paolo]

Hello,

this code compiles and works. Please note the space character around dots and after "add", is this behaviour correct ?

Code: Pascal  [Select][+][-]

1. procedure TForm1.Button1Click(Sender: TObject);
2. begin
3.   ListBox1 . Items . Add ('x');
4. end;
5.  

[Fibonacci]

Yes, any white characters

[TRon]

Yes, any white characters

Please elaborate on the source for that answer if possible.

From a technical point of view it is wrong.

[Fibonacci]

From a technical point of view it is wrong.

Why? Sometimes I use it like this:

Code: Pascal  [Select][+][-]

1. something
2. .add(0)
3. .add(1)
4. .add(2)
5. .execute;

If new lines are allowed, then why spaces or any other white character should not.

[TRon]

If new lines are allowed, then why spaces or any other white character should not.

Because a (white) space is not valid for an identifier. Ofc as can been seen the compiler simply ignores them but that still does not make it right. That is why I asked for the source. If it isn't documented it is undefined behaviour and as such should not be used (hence me asking) 

[Fibonacci]

That is why I asked for the source.

Experience

[TRon]

Experience

Ok, got it. thx.

[440bx]

That behavior isn't usually explicitly documented but it is a consequence of the tokenizing and production definitions.  IOW, extremely likely that all Pascal compilers (that support selection using a ".") support that construction.

It's not just whitespace but any "null token", for instance:

Code: Pascal  [Select][+][-]

1. type
2.   ARecord = record
3.     Field1    : integer;
4.   end;
5.  
6. var
7.   v : ARecord;
8.  
9. begin                        
10.   v { a comment before the field selector } . { another comment } Field1 := { comment } 1 { comment }; { comment }
11. end.
12.  

You can have block comments where whitespace is allowed.  The same thing can be done with other constructions, e.g., "array1 { comment } [ { comment } 1 { comment } ] := ...".  IOW, any token that is a delimiter is also a separator therefore a { comment } can usually (probably always} precede or follow it.

It's documented in the way Pascal is tokenized and by the structure common to all its productions.  This is part of what makes Pascal a free form language (unlike, for instance, Python).  Same thing can be done in C/C++ and other free form languages, again for the same reason.

[TRon]

Thank you for the elaboration 440bx.

I have located my line of thought leading up to my response. I thought it was forbidden but I mixed that up with the Delphi style-guide in which it is not recommended to add white-space(s) around dots.

[440bx]

Thank you for the elaboration 440bx.

You're welcome TRon, my pleasure.

[n7800]

Just as you put spaces at the beginning of a line for indentation, they can also be between identifiers/operators, since they are not part of it:

Code: Pascal  [Select][+][-]

1.  
2. // different number of spaces between operators
3. a:=b+c-d;
4. a := b + c - d;
5.  
6. // connecting short operators on one line
7. if true then
8.   exit;
9. if true then exit;
10.  
11. // splitting a long operator into different lines
12. if Function1(Variable1, Variable2, Variable3,
13.   Variable4, Variable5, Variable6) = Variable7
14. then
15.   exit;
16.  

[n7800]

In practice, this is often used for "horizontal alignment" of similar lines:

Code: Pascal  [Select][+][-]

1. // without alignment
2. test.one:='value1';
3. test.two:='value2';
4. test.three:='value3';
5. test.four:='value4';
6.  
7. // with alignment
8. test.one   := 'value1';
9. test.two   := 'value2';
10. test.three := 'value3';
11. test.four  := 'value4';
12.  

This usually makes it easier to read and edit, if you don't overdo it.

[n7800]

In general, if you are so surprised by the possibility of adding spaces, then the possibility of their absence will stun you:

https://gitlab.com/freepascal.org/fpc/source/-/issues/40967

[440bx]

In general, if you are so surprised by the possibility of adding spaces, then the possibility of their absence will stun you:

https://gitlab.com/freepascal.org/fpc/source/-/issues/40967

Yes, another fun one (among a large number of them) is:

Code: Pascal  [Select][+][-]

1.   for i:=1downto-2do;
2.  

Delphi accepts the above and FPC accepts the above and also

Code: Pascal  [Select][+][-]

1.   for&i:=1downto-2do;

It would be fairly easy to modify the scanner to require whitespace after some tokens but, it could be a bit annoying.

[MarkMLl]

It would be fairly easy to modify the scanner to require whitespace after some tokens but, it could be a bit annoying.

Things get messy, particularly with operators such as mod and div positioned adjacent to a variable (i.e. no interposed whitespace, parenthesis etc.) and with & indicating an octal number.

Code: Pascal  [Select][+][-]

1. if i&mod&7 = 0 then
2. ...
3.  

MarkMLl