Announcing: port_upgrade

Scott Haneda talklists at
Sat May 9 18:35:22 PDT 2009

Why can't the wireshark port and similar ports have a "requires- 
depends-rebuild" defaulting to false.

The port file could set it to true as needed. Some other "magic" could  
happen in the mac ports base to figure out if it needs special  

If no other ports are registered to depend on the parent dependency,  
it could just deactivate, uninstall, and install.

I see there being issue if the main ports dependency has other port  
installed software that depends on the same software.

In those cases an interactive question could be asked.

I've actually been meaning to post here asking if making mac ports  
have interactive modes would be a good idea.

My gut tells me a large amount of the questions on this list end up  
being answered with instructions to add a -f or some other semi  
agressive flag or set of forced uninstall reinstall steps.

Could this not be solved with yes/no interaction or a/b/c style q&a in  
the shell?

I could see a very simple set of base conditions one could add to a  
port file. Ports could return codes for every error and port  
developers could have a base of directives to use to throw the install  
into an interactive mode.

Just some food for thought. At the very least it may afford the  
removal of a few wiki pages and free the list up from reprtitios  

Also wanted to suggest "progress bars" for operations. I used yum the  
other day and their little dashed line progress bar was pretty nice.
Iphone says hello.

On May 9, 2009, at 5:32 PM, "Scott C. Kennedy" <sck at> wrote:

> Okay, maybe  I'm being a bit stupid here, but if I had I compiled  
> wireshark several months ago, and it required libpcap, however the  
> port for wireshark was upgraded BEFORE the port for libpcap... So,  
> if I don't rebuild wireshark AFTER the libpcap upgrade I would have  
> a problem like "The wireshark package depends on the major version  
> of libpcap, and as such requires rebuild when a new major version of  
> that package is pushed. "
> But if I use port_upgrade, then I don't in fact jmr in this ticket  
> even agreed that "If someone uses "port upgrade outdated" and the  
> bumped libpcap doesn't result in dependencies (like wireshark) being  
> rebuilt, I think that's a bug with "port upgrade" and should not be  
> fixed by band-aiding ports every time their dependencies are  
> changed. "
> So, now there is an option for those of us wanting tools to work  
> reliably, and don't wish to spend hours trying to determine which  
> library version of hat dependancy was outdated by what to cause a  
> bug, when a simple re-compile solves the issue the first time around.
> I fail to see the issue.
> Scott
> Bryan Blackburn wrote:
>> On Sat, May 09, 2009 at 04:28:46PM -0700, Scott C. Kennedy said:
>>> Plus, when a dependency for a package changed significantly, then  
>>> the
>>> program would terminate or act unexpectedly. For me, a longer  
>>> compile
>>> time when updating packages to ensure that all my packages are  
>>> linked to
>>> the most recent version of all of their dependencies is a good
>>> investment of time.
>> 'port upgrade' when new dependencies are added should simply  
>> install those
>> ports, what is the problem you've seen, is there a ticket?
>> Bryan
> _______________________________________________
> macports-users mailing list
> macports-users at
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the macports-users mailing list