During this morning's hackfest we discussed and agreed to a tentative road-map for armv5, v6, v7, and v8. Part of this is in the context of moving build infrastucture to the PHX colocation facility. Part of this is in the context of pushing ARM to Primary architecture status.
The following is a summary of what we agreed to:
Any release we do as part of Fedora ARM will be supported according to PA life cycle (IE, end of life between PA and SA ARM release will be the same day).
Any release we do as a secondary architecture will continue to be supported by the secondary architecture koji instance, even if later releases are done as a primary architecture. For example, F18-ARM will always be secondary because it was built on the secondary koji hub.
In a couple weeks' time we will have 96 Calxeda Highbank servers in PHX. There is already a new koji hub and database in PHX ready to take over. Once the ARM systems are running smoothly we will migrate armv5tel and armv7hl builds to these machines. Dennis Gilmore will take the lead in migrating koji to the new systems.
Here is the plan on an architecture-by-architecture basis:
armv5: F19 will be the last release we build on armv5tel. We will continue to support F19 armv5tel throughout the supported lifecycle of F19. This means packages in f19-testing, f19-updates, etc will continue until they're shut down in prmiary.
armv6: Seneca will continue their development of armv6hl on the hardware at Seneca college. The systems currently working on armv5tel and armvhl will be redirected to this effort once those architectures are migrated to PHX- allowing them to make faster progress.
armv7: The goal is to push armv7hl as a primary architecture in the Fedora 20 or 21 timeframe. Once accepted some of the builders in PHX will be moved to the primary network infrastructure, but a sufficient number will be left as secondary builders to continue support for F19 and earlier throughout their lifecycles.
armv8: The aarch64 bootstrap effort continues to progress and is now at the early portion of stage 3: Rebuilding packages natively with rpmbuild. To follow will be a prolonged stage 4 in which most packages are built with mock. When we bootstrapped armv7hl stage 4 was a significant group effort and will hopefully be so again. It is likely we will reach stage stage 4 before v8 hardware is generally available, so every person's assistance will be especially welcome. The hope is to have v8 support in koji as a secondary architecture shortly after hardware becomes available. Our educated guess is that we will be somewhere between F20 and F21 development when v8 is ready to be guided by koji as an official secondary architecture.
That's it!