From: Beman Dawes (bdawes_at_[hidden])
Date: 2002-08-14 10:59:46
At 11:36 AM 8/14/2002, Douglas Paul Gregor wrote:
>On Wed, 14 Aug 2002, William E. Kempf wrote:
>> > > * !!(predicate) fix as discussed
>> >
>> > No longer necessary. I switched to BOOST_TEST(p? true: false); adding
>!!
>> in
>> > BOOST_TEST would silently test operator!, not the "boolean context"
>> > conversion.
>>
>> I'd say it would still be necessary since other people are sure to use
>the
>> same construct. However, you bring up a good point about it causing
the
>> wrong thing to be tested, so !! is the wrong solution. Probably need
to
>> apply the ternary operator instead.
>
>Maybe we can introduce another macro here, such as
>BOOST_TEST_BOOLEAN_CONTEXT, so that what we are checking is more
explicit?
>I'm getting big by this same issue in Function, and x? true : false is,
>IMHO, quite unclear
Ouch! I hate to see that kind of alternative behavior based on a macro.
For years I've avoided implicit conversions to bool, which usually means
adding "== 0" or "!= 0", but sometimes "? ... : ..."
I started doing that originally to avoid compiler warnings, but over time
came to view the resulting code as clearer and more explicit when viewed by
a maintenance programmer. It did take some getting used to, however, but
now feels totally natural.
--Beman