odd block indentation

[kjx]

valueOf { print(1)
                              print(2)
                              print(3) }  

prints

1
2
3

shouldn't it complain about a missing print(_)print(_)print(_) method?

[apblack]

I don't think so.

We have to decide if the increase in indentation on line 2 is to indicate a continuation line, or because a block has started. Because line 1 has an unmatched {, line 2 starts a block, and the indentation is assumed to be that required by starting a block.

If instead you write

valueOf { print(1)
print(2)
print(3) }  

then you get Syntax error: Please indent the body of a block.

In hindsight, it may have been a mistake to use indentation for these two disparate purposes. But the escaped newline convention used by Python (and shell) is so ugly ...

[kjx]

Yeah escaped newlines are surely evil.
Given that indentation of print(1) starts at col 11, I don't see why when it comes to the indentation of print(2):

• anything to the left of column 11 is an error
• at column 11, a new separate statement
• anything to the right of column 11 is a continuation.

that's to say the required indentation inside a code sequence is the indentation of the start of the first statement in the code sequence, where than needs to be to the right of the previous indentation.

[apblack]

Given that indentation of print(1) starts at col 11,

Grace is not Haskell. We count not the indentation of a statement; rather we count the indentation of the line that contains the statement. Hence, thou shalt say

the indentation of the line containing `print(1)` is zero

It shalt also also be seen that the line containing print(1) openenth a block, as the { bears withness thereto

Hence, until the closing }

• every statement shall be indented 2 or more

It is not possible to both start a block on the current line, that is, to put a statement after an unmatched opening {, and to continue that statement on the next line. If the statement won't all fit to the right of the {, start a new line after the {

[apblack]

Grace is not Haskell

Which is why most of the Haskell-inspired work on indentation inference and specification was not much help to us.