On Wed, Oct 12, 2005 at 07:00:32PM +1000, Finn Thain wrote: > If you take the long view, and assume that we will get prefixes sooner > than later, then devs should be aiming for _maximum_ collisions, since > from a darwin point of view, that means better interoperability with > Apple's open source work.
That simply requires people to have machines for this. I don't have machines floating around, and I'm fore sure not willing to buy hardware or make my box into an even more unstable thing than it is already right now (try to attach an iPod to it, and it completely goes nuts for examplei, yeah, it's just Windows(tm)(r)). From a managerial point of view, your note sounds correct to me, but considering the resources, it's not simply done. Unless after this mail two or three people step forward that are willing to do this maximum collision stuff. > If you take a compromise, you might end up with fewer collisions in the > short term, but you make it harder for Gentoo/Darwin and "progressive" to > interoperate with Gentoo/macos and Apple. We cannot control collisions, at least not the real ones. They are there and doing something about it, usually results in having more work to do once we get a prefix. > That is why I argued against moving the perl executable, for example. And > it is also why I argued for stabling packages with collisions. I was > simply taking the long view, and trying to avoid rework for the > gentoo/darwin project. I don't like to solve perl for the current situation. It is typically something that should be dealt with in a prefix. If someone wants to install it on a progressive box, please do so. > As for the "conservative" profile, it doesn't have many users, and will > not have until we get prefixes, so why optimise for "collision-protect"? Given the reponses and signals I see, there are even less progressive users, so why optimise for them for now and screw my box to try and stable those ebuilds? If someone else wants to do it, fine with me, but it simply seems not to be an option, as noone wants to do this. I simply only want to keyword those things that for instance Dirk now points out that already works, and probably also will work in a prefix. You can call it lazy or short sighted, but to me it's just a matter of resources, and especially hardware resources I don't have. If I'd have access to a machine I could screw by installing for instance GCC via portage, then I'd try it out to see if it compiles and eventually keyword and mask it. I simply don't have it! -- Fabian Groffen Gentoo for Mac OS X Project -- Interim Lead -- [email protected] mailing list
