Re: Questions about behavior observed in gm2

[Gaius Mulley]

Jack Perry <devotus@yahoo.com> writes:

Hi Jack,

> Sure. I don't know what the language standard is, but as to the other
> John's question: from my point of view, the bug is not that one has to
> write Part1() with parentheses, but that omitting the parentheses was
> not a compiler bug. Instead, it did a silent c-style cast, without even
> raising a warning. My impression was that this is at least a type
> mismatch: `Part1` is a function, not an integer, so if you want the
> function's address, you should be explicit about that.

now fixed when using -fiso (and git pushed).  The pim dialects still
accepts the code and the pim fix is still to be implemented.

> If I may, I'd like to ask about another thing I observed the other day. I 
> have a variant record
>
> ParsedDigit = RECORD
> CASE Valid: BOOLEAN OF
> TRUE:
> Value: CARDINAL;
> ELSE
> END;
> END;
>
> Suppose I want to initialize it, e.g.,
>
> CONST PD1 = ParsedDigit { False } ;
> PD2 = ParsedDigit { True, 3 };
>
> FWIW this is not the particular one I was having trouble with; but the
> compiler refused to accept it. It kept saying I had an unknown
> field. Is this a known issue?

yes indeed this appears to be a bug - thanks for the bug report - will
fix and add it to bugzilla for tracking

regards,
Gaius