Question about build dependencies

[Joshua Root]

Ryan Schmidt wrote:
> On May 29, 2008, at 07:43, Tabitha McNerney wrote:
>> For example, it might be a good idea to try and have as many  
>> dependencies for ports (whether build, run or library) be based on  
>> other ports rather than what Apple provides on the root filesystem  
>> because Apple can (and will) change their mind about things (which  
>> is fine with me just as long as for most if not all dependencies I  
>> can use MacPorts and not depend on Apple's frame of mind)!
> 
> Which is precisely why the MacPorts project recommends that ports  
> depend on other ports, not on binaries or libraries that may be  
> installed on the system by Apple, whose versions or features could  
> change under our noses by a system update.
> 
> http://trac.macports.org/wiki/FAQ#WhyisMacPortsusingitsownlibraries
> 
> I'm not sure why the libiconv port was written differently here.  
> Looks like Mac OS X 10.4 provides gperf 3.0.1. MacPorts provides  
> gperf 3.0.3. Maybe I should change the port to use the gperf port  
> exclusively. It would be more consistent with our policy.

Back in the olden days of DarwinPorts, there were no such thing as port: 
dependencies. (They were added in 2005, r11790 if you're curious.) The 
libiconv port predates that, and I thought that might be why it uses bin:.

However, it turns out that the gperf dependency was actually first added 
in r21556 in early 2007. Ironically, the exact issue of inconsistent 
features in the Apple-provided gperf was encountered shortly thereafter, 
and a port:gperf dependency was added, for darwin 7 only, in r21795.

So yes, the dependency should probably be changed to port:gperf for all 
platforms. :-)

- Josh