svn commit: r236377 - head/sys/dev/vxge/vxgehal

[John Baldwin]

On Friday, June 01, 2012 12:39:48 pm Eitan Adler wrote:
> On 1 June 2012 07:24, John Baldwin <jhb at freebsd.org> wrote:
> > This is why I personally loathe assignment side effects in boolean expressions
> > for control flow.  I tend to write this sort of thing instead as:
> >
> >        channel->dtr_arr[dtr_index].dtr = dtrh;
> >        if (dtrh != NULL) {
> 
> Same here. I was told to use the assignment is style(9) by multiple
> people though. If it really is, I wish it would change.

style(9) doesn't make a clear statement either way, but it has contradicting
examples.

First:

             while ((ch = getopt(argc, argv, "abNn:")) != -1)

(and this one I use all the time myself as that is the common idiom for
getopt())

Second:

             error = function(a1, a2);
             if (error != 0)
                     exit(error);

Also, style(9) tends to frown on assignments that are side-effects in other
places, for example:

     Be careful to not obfuscate the code by initializing variables in the
     declarations.  Use this feature only thoughtfully.  DO NOT use function
     calls in initializers.

             struct foo one, *two;
             double three;
             int *four, five;
             char *six, seven, eight, nine, ten, eleven, twelve;

             four = myfunction();

Some newer changes added at the end do use assignment side-effects while
demonstrating other rules (the !p example and err(3) and warn(3) examples,
this last I find particularly obfuscated and painful to read).

Note that I do not consider this to be an assignment side effect:

	if (ioctl(...) < 0)
		err(1, "ioctl");

That is, I don't write that out as:

	i = ioctl(...);
	if (i < 0)

However, I do split it out if the return value is used for more than error
checking, e.g.:

	fd = open(...);
	if (fd < 0)
		err(1, "open");

-- 
John Baldwin