svn commit: r305357 - head/net/ssvnc

[Chris Rees]

On 8 October 2012 08:16, Baptiste Daroussin <bapt at freebsd.org> wrote:
> On Mon, Oct 08, 2012 at 04:59:06AM +0000, Alexey Dokuchaev wrote:
>> On Mon, Oct 08, 2012 at 04:01:54AM +0000, Alexey Dokuchaev wrote:
>> > On Sat, Oct 06, 2012 at 11:23:17AM +0000, Baptiste Daroussin wrote:
>> > > New Revision: 305357
>> > > URL: http://svn.freebsd.org/changeset/ports/305357
>> > >
>> > > Log:
>> > >   fix sense of a test
>> > >
>> > > -.if ${PORT_OPTIONS:MULTRAFTP}
>> > > +.if empty(PORT_OPTIONS:MULTRAFTP)
>> >
>> > That looks weird, and commit message is pretty cryptic as well.  Could you
>> > explain what went wrong here so others would not fall in the same pitfall?
>>
>> Err, I read is as ! vs. empty() test.  Still, I've noticved some arguments
>> on the syntax, would be nice to know if ! ${PORT_OPTIONS:M...} has issues
>> other than purely stylish.
>>
>
> No it hasn't (as far as I know)
>

This is all my fault-- I was the one who suggested the original use of
${FOO:M}, and I deliberately chose to retain empty(FOO) as the reverse
because I was convinced that the insanity of make's conditionals would
screw us over in some way.

I have at [1] pulled together some simple tests, and I think I've
caught all use cases here.

It seems to work fine-- are there any people more wizardly than me
with make that can confirm that it's OK?  If so, I'll update the docs.

Chris

[1] http://www.bayofrum.net/~crees/scratch/Makefile-do-we-need-empty-after-all