On 5 September 2016 at 17:35, Nicholas D Steeves <nstee...@gmail.com> wrote: > On Tue, Jul 26, 2016 at 12:35:56PM +0100, Dimitri John Ledkov wrote: >> On 20 May 2016 at 22:48, Nicholas D Steeves <nstee...@gmail.com> wrote: >> > >> > Because btrfs-convert is known to fail, and is known to causes >> > data-loss that may take some months to manifest, we must provide a >> > warning, because data-loss is serious. This patch also includes a >> > contrib initramfs hook which is installed to >> > /usr/share/doc/btrfs-progs; it must be manually copied into place; >> > I've included it because this is the first place a sophisticated user >> > would look. >> > >> >> Adding a btrfs-convert initramfs hook (albeit disabled) and warning >> not to use it, sounds a lot like negative suggestion encouraging >> people to use it. >> >> I do not want to ship btrfs-convert (nor provide suggestions) in the >> initramfs precisely because upstream discourages people from using it. >> >> I also fail to see how this is an RC bug. btrfs-convert is a crutch >> for people to toy and try out btrfs, and always has been. > > Primary issue: btrfs-convert provides no warning > Secondary issue introduced by proposed fix: btrfs-convert is a diversion > script > > To solve the primary issue, would you prefer a quilt patch to the source? > The problem of encouraging masochistic users is only relevant to the > workaround for the issue a diversion script creates. > > Raid5/6 profiles are also now on the list of features upstream considers > dangerous, but there is yet no warning. >
Please submit warnings as patches upstream -- Regards, Dimitri.