On 7/17/2012 20:47, John Nemeth wrote:
On Dec 7, 4:26am, John Marino wrote:
}
} So reverting the commit is punishment?
} Because documented in PLIST or not, that file is getting installed on
} DragonFly. When it's understood "why chan_mgcp is built only in some
} situations", then an alternative solution can be added. I don't
} understand why one would remove a working solution without something
} ready to replace it.
It's not a working solution. As you've been told before, when you
see an issue on Dragonfly, it quite likely exists elsewhere as well.
Making OS specific hacks is most likely wrong. Most likely there
should be some kind of feature test. And, yes, removing unapproved
gross hacks is appropriate.
}-- End of excerpt from John Marino
I've replied in-depth offline, but for public benefit: I don't agree about the timeframe of removing the hack.Let's say there are 3 platforms out that that you were previously unaware were generating this library, DragonFly being one. After the "hack", there are only two broken platforms. The introduction of the (unapproved) hack informs you of a problem.
You could:A) remove the hack now, going back to 3 broken platforms until a real fix comes along (sometime in the indefinite future, maybe) B) At your leisure, come up with a real fix, implement it and remove the hack.
Option A is better than B. The other two platforms would be no worse off with either option, and one would enjoy correctly built asterisk packages.
I think that horse is dead now. John