in reply to Re^2: Complex conditional sort
in thread Complex conditional sort

Fair enough ikegami.

I like using the !!$test construct to force a boolean to 1 or (''/0). However, if that isn't clear enough, I'd probably choose using the >0 test.

sort {$products{Cat}{Pro}{$b}{Qua}>0 <=> $products{Cat}{Pro}{$a}{Q +ua}>0 || $a <=> $b}

I suppose it still requires some level of understanding of the possible return values of conditionals, but that's one of the most important things to be familiar with when dealing with sorting in my opinion. Especially the <=> and cmp operators.

Replies are listed 'Best First'.
Re^4: Complex conditional sort
by ikegami (Patriarch) on Mar 09, 2011 at 07:15 UTC

    but that's one of the most important things to be familiar with when dealing with sorting in my opinion. Especially the <=> and cmp operators.

    Only those. Again, Perl makes no guarantee as to what the others return. It's definitely not important to be familiar with what they return as they can return anything true and anything false. In fact, they don't always return the values you think they return.

    I'd probably choose using the >0 test.

    Same exact problem as negation.

      I figured you'd say that. I hear ya when it comes to perl not "promising" any particular return value. But can you name any example where !!$test doesn't return 1 or ''/0 currently? And do you really think it's likely that's going to change?

      Obviously, for safety sake with future versions, it probably is better to rely on documented return values instead of assumed current implementation. However, it seems a little silly to me to be that wary of something so basic changing. I wouldn't have been the only person to occasionally take advantage of observed behavior instead of only relying on documented. To future ill or not.

        But can you name any example where !!$test doesn't return 1 or ''/0 currently?

        Overloaded operators can do this.* I'm talking about properly written overloaded operators, not ones that try to repurpose "!". It's not an issue here and it's not likely to ever be an issue, which is why that's not the argument I chose to make.

        What I did focus on is the writing of unreadable code to save one character. I find it too demanding of your reader to know that "!" returns +1 for true given the lack of cost of not demanding this knowledge.

        The cost of confusion is low, but so is the cost of avoiding it.

        * —