Re: Parsing/evaluating shortcutting operators

Jeremy J Starcher <>
Sat, 17 Jan 2009 19:39:56 GMT

          From comp.compilers

Related articles
Parsing/evaluating shortcutting operators (Jeremy J Starcher) (2009-01-13)
Re: Parsing/evaluating shortcutting operators (Bartc) (2009-01-15)
Re: Parsing/evaluating shortcutting operators (Jeremy J Starcher) (2009-01-17)
| List of all articles for this month |

From: Jeremy J Starcher <>
Newsgroups: comp.compilers
Date: Sat, 17 Jan 2009 19:39:56 GMT
Organization: at&t
References: 09-01-032 09-01-035
Keywords: code
Posted-Date: 18 Jan 2009 08:06:34 EST

On Thu, 15 Jan 2009 18:05:08 +0000, Bartc wrote:
> "Jeremy J Starcher" <> wrote in message
>> I am unsure how to proceed in setting up shortcutting operators
>> however. For example, this code:
>> declare boolean b
>> b = (1 == 2) || (3 == 4)
>> Which results in the expression
>> 1 2 == 3 4 == || b =
> You've made me feel inadequate, because my compiler only uses shortcut
> evaluation when there are jumps involved (eg. inside an If conditional,
> but not in an ordinary expression).

Hey.. don't feel that way. You are a step over where I am.

> But I found that my (a|b|c) construct (equivalent to C's a?b:c)
> generates code like an If statement.
> So I can fix this 'problem' by turning 'a and b' into (a and b|1|0), and
> 'a or b' into (a or b|1|0), either manually or letting the compile do it
> as you seem to be attempting to.

Interesting idea .... Hmmmm... Room to think. Thank you.

> It's possible that you can try something similar, if you also have an
> equivalent a?b:c operator that already has short-circuit evaluation.

And here I was, going to make the trinary operator work via the
shortcutting operators ;) Chicken-and-egg.

To get it, I think I'm going to have to come up with a different equation
parser, and split the calculations into 'sub-expressions'. Not that I'm
ever going to add optimization, but it would also 'common-subexpression'
elimination, as well as folding constants.

Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.