Question about style

[spibou]

What's your opinion on something like "x += x < MAX_VALUE" where
x is an integral type ? Elegant or ugly and incomprehensible ?

 

[Mike S]

What's your opinion on something like "x += x < MAX_VALUE" where
x is an integral type ? Elegant or ugly and incomprehensible ?

Both ;-)

IMHO the "correctness" of a particular coding style can't be judged
objectively; there will always be personal bias as to which way is The
Right Way. Ultimately, it's up to you to decide whether you can read it
and whether it's acceptable or not, unless you're modifying someone
else's code, in which case it's better to adopt their coding style for
the sake of consistency and clarity.

That being said, I'm not sure I like the specific example you gave.
It's mildly clever, but I would prefer to see something like

if (x < MAX_VALUE) ++x;

because it makes the intent much clearer.

 

[spibou]

Both ;-)

IMHO the "correctness" of a particular coding style can't be judged
objectively; there will always be personal bias as to which way is The
Right Way.

Perhaps someone will find an objective way to evaluate it. Even if
not I'd still like to have a collection of subjective opinions.

That being said, I'm not sure I like the specific example you gave.
It's mildly clever, but I would prefer to see something like

if (x < MAX_VALUE) ++x;

That is indeed the most obvious way of doing it. The problem with
it is that you can only use it in places where the language allows a
statement as opposed to an expression. Of course you can also use
x += x<MAX_VALUE ? 1 : 0 but I find this somewhat silly since
x<MAX_VALUE taken on its own already has the correct values.

Personally I'd use either one depending among other factors on the
whim of the moment. And I have to say that I don't use it for
reasons of cleverness. I have simply developed a taste for
expressions of this sort where the numerical value of a logical
expression is used for numerical computations since my early BASIC
days because I read on some magazine back then that the interpreter
is likely to execute these faster than using if.

 

[Mike S]

[snip]

It's mildly clever, but I would prefer to see something like

if (x < MAX_VALUE) ++x;

[snip]

Personally I'd use either one depending among other factors on the
whim of the moment. And I have to say that I don't use it for
reasons of cleverness. I have simply developed a taste for
expressions of this sort where the numerical value of a logical
expression is used for numerical computations since my early BASIC
days because I read on some magazine back then that the interpreter
is likely to execute these faster than using if.

Well, to be honest, the expression form x += x < MAX_VALUE is already
starting to grow on me; I would use parentheses though:

x += (x < MAX_VALUE);

(In fact, I when I first retyped the expression, I unconsciously
inserted the parentheses, I guess out of habit...). Anyway, I didn't
mean to make it sound as if using an "if" statement would be better or
worse - an "if" would simply serve to make your intention (painfully)
obvious. There is at least some merit in this as a general practice
when writing code, for the benefit of people reading your code, but at
the same time, if e.g. a future maintenance progammer can't untangle
the meaning of x += (x < MAX_VALUE) after a few seconds, then they
aren't qualified enough to maintain *anyone's* code ;-)

 

[Arild Hystad]

What's your opinion on something like "x += x < MAX_VALUE" where
x is an integral type ? Elegant or ugly and incomprehensible ?

Personally, I would have used parantheses to make my intention clear, like
this: "x += (x < MAX_VALUE)". It could also be written like this:
"(x < MAX_VALUE ? ++x : x)".

 

[lovecreatesbeauty]

What's your opinion on something like "x += x < MAX_VALUE" where
x is an integral type ? Elegant or ugly and incomprehensible ?

It's ok. The expression is same as:
y = x < MAX_VALUE;
x += y;

The operator '<' takes precedence over '=' in evaluation. I think it's
good to make the equality clear and nature. But something like "x++ +=
x++ < MAX_VALUE" is a rather bad idea.

 

[Keith Thompson]

It's ok. The expression is same as:
y = x < MAX_VALUE;
x += y;

The operator '<' takes precedence over '=' in evaluation. I think
it's good to make the equality clear and nature. But something like
"x++ += x++ < MAX_VALUE" is a rather bad idea.

It's not just a bad idea, it's illegal (more precisely, it's a
constraint violation). "x++" does not yield an lvalue, so it can't
appear on the left side of an assignment. (Even if that weren't the
case, of course, modifying x twice between sequence points would
invoke undefined behavior.)

 

[Tom St Denis]

What's your opinion on something like "x += x < MAX_VALUE" where
x is an integral type ? Elegant or ugly and incomprehensible ?

Use a fscking if stmt. You'll thank me when your first project hits
100K lines, hundreds of files and you can STILL READ THE DAMN CODE.

if (x < MAX_VALUE) { ++x; }

That may be longer [I would indent and put the ++x; on a newline] but
it's immensely easier to read [at high speed] and will prevent you from
ripping your brain out of your skull when you have to maintain 100K
lines of gibberish code.

Tom

 

[lovecreatesbeauty]

It's not just a bad idea, it's illegal (more precisely, it's a
constraint violation). "x++" does not yield an lvalue, so it can't
appear on the left side of an assignment.

Thank you for the reminder. Yes, it will fail at compilation. And more
accurate, `x++' doesn't yield a modified lvalue

(Even if that weren't the
case, of course, modifying x twice between sequence points would
invoke undefined behavior.)

Thanks for the reminder again. I was going to express myself on this
point in my previous post but I failed. You know me

lovecreatesbeauty

 

[Malcolm]

What's your opinion on something like "x += x < MAX_VALUE" where
x is an integral type ? Elegant or ugly and incomprehensible ?

Compileable gibberish.

At very worst make it x += (x < MAX_VALUE) ? 1 : 0;

 

[CBFalconer]

Compileable gibberish.

At very worst make it x += (x < MAX_VALUE) ? 1 : 0;

It's already obscured, why obscure it further with the useless ?:
operator? All that can do is permit the compiler to generate
useless code.