Handling the movement of a header file between packages

[Ryan Schmidt]

On Dec 28, 2011, at 20:13, Jeremy Lavergne wrote:

>>> The way we have handled this in other ports before is to use the "deactivate hack": At pre-activate, you test whether the header already exists, and if so, if a version of that other port is active that is known to have provided the header; if so, you deactivate that port. See examples here:
>>> 
>>> https://trac.macports.org/ticket/28865#comment:3
>> 
>> Seems to be just what I'm after, thanks.
> 
> If we provided an official way of doing this, we could recover successfully if the user cancels the action while wer're in the middle of deactivating a separate port like this.

Canceling in the middle of activating or deactivating a port is probably always bad, regardless of whether it's as a result of this kind of code or not.