So what do you guys think of leaving behind empty stubs for compatibility and then simply filing a tracking bug blocked by any packages that removing herds broke?
On Sat, Feb 13, 2016 at 2:44 PM, Rich Freeman <[email protected]> wrote: > On Sat, Feb 13, 2016 at 4:47 PM, Raymond Jennings <[email protected]> > wrote: > > The guys making the API change bear the burden of fixing anything it > breaks, > > however, if something gets officially deprecated, don't go out of your > way > > to support continued use. > > We tend to do this already for things like PMS, which is as close as > Gentoo gets to something like the kernel API. > > However, sometimes a gradual transition doesn't always make as much > sense, and Gentoo doesn't always have the manpower to make every > change a pretty one. > > And there is a cost to maintaining that kind of backwards > compatibility. For example, debian chose to keep its LSB init scripts > and write systemd unit wrappers around them. If they had chosen > openrc instead I wouldn't be surprised if they kept the LSB init > scripts and wrote an openrc compatibility layer around that. While > this does provide a more stable experience, it also leaves around a > ton of cruft. > > In general I tend to favor a balance. Trying to get everything just > right was why the git migration literally took years, and even that in > the end had a few bumps. Gentoo users need to be willing to deal with > the occasionally bump in the road - we try to provide a fairly cutting > edge experience, with minimal layers in integration. > > But, there is nothing really wrong with your suggestion, and we try to > accommodate that approach when we can. > > > And yes, the personal attacks probably should die down. > > ++ > > -- > Rich > >
