bug#33744: 26.1; ada-mode 6.0.0 indentation of operators starting a line in a multi-line expression

bug#33744: 26.1; ada-mode 6.0.0 indentation of operators starting a line in a multi-line expression

[Ludovic Brenta]

Package: emacs, ada-mode
X-Debbug-CC: ada-mode-users@nongnu.org

Hello, consider the line marked "incorrectly indented" below:

procedure Operator_Indentation is
begin
   if B
        or else C
        > 2 -- incorrectly indented
   then
     null;
   end if;
end Operator_Indentation;

We think the indentation performed by ada-mode 6.0.0 is incorrect.
Of course we understand that the line is indented relative to B
but as this line is part of a sub-expression, we think it should be
indented relative to the sub-expression instead.  We cannot decide
which of the two following alternatives is the most "correct":

procedure Operator_Indentation is
begin
   if B
        or else C
          > 2 -- indented relative to "or else"
   then
     null;
   end if;
end Operator_Indentation;

procedure Operator_Indentation is
begin
   if B
        or else C
                  > 2 -- indented relative to C, which starts the 
sub-expression
   then
     null;
   end if;
end Operator_Indentation;

(we set ada-indent to 2).

Of course, these examples are overly simplified; in our real code base,
we don't write our operators on a separate line unless the previous line
("C" in these examples) is very long.  To illustrate this, our actual 
code
looks more like:

procedure Operator_Indentation is
begin
   if Blarg_Meets_Preconditions_For_Subsequent_Test (Blarg)
        or else Critical_Cruising_Configuration_Condition_Code 
(Blarg.Blurp)
          > 2
   then
     null;
   end if;
end Operator_Indentation;

PS. The value of ada-indent-hanging-rel-exp has no effect in this
particular case.

-- 
Ludovic Brenta.

bug#33744: effect of ada-indent-hanging-rel-exp

[Stephen Leake]

Ludovic writes:

> The value of ada-indent-hanging-rel-exp has no effect in this
> particular case.

Actually, it does. With ada-indent = 2, ada-indent-hanging-rel-exp =
nil, the indentation is:

procedure Operator_Indentation is
begin
  if B
    or else C
    > 2 -- indented relative to "or else"
  then
    null;
  end if;
end Operator_Indentation;

Changing an indentation parameter does not force a reparse, so calling
'indent-region' does not use the new parameter value. Use M-x
wisi-parse-buffer to force a reparse and reindent.

-- 
-- Stephe

bug#33744: effect of ada-indent-hanging-rel-exp

[Stephen Leake]

Ludovic writes:

> The value of ada-indent-hanging-rel-exp has no effect in this
> particular case.

Actually, it does. With ada-indent = 2, ada-indent-hanging-rel-exp =
nil, the indentation is:

procedure Operator_Indentation is
begin
  if B
    or else C
    > 2 -- indented relative to "or else"
  then
    null;
  end if;
end Operator_Indentation;

Changing an indentation parameter does not force a reparse, so calling
'indent-region' does not use the new parameter value. Use M-x
wisi-parse-buffer to force a reparse and reindent.

-- 
-- Stephe

bug#33744: effect of ada-indent-hanging-rel-exp

[Stephen Leake]

Is the fact that the first token on a line is an operator significant?
In other words, would this indentation be correct:

procedure Operator_Indentation is
begin
  if B or else
       C >
         2
  then
    null;
  end if;
end Operator_Indentation;

-- 
-- Stephe

bug#33744: 26.1; ada-mode 6.0.0 indentation of operators starting a line in a multi-line expression

[Ludovic Brenta]

Hello,

Our coding standard forbids operators at end of line, so in our case
we would always place the operator at the beginning of the next line.
Were it not for that coding standard, I think your example would be
correct.  But this does not solve our problem :)

-- 
Ludovic Brenta.