[llvm-commits] Committing AArch64?

[Tim Northover]

Hi all,

I've uploaded new versions of the AArch64 patches today. I believe the changes 
are minimal this week (ABI generation refactoring due to ToT work in Clang; 
foreach loop for registers; and fixing a layering violation in Disassembler).

I'd like to move towards committing it as an experimental target as soon as is 
reasonable. I believe the risk for existing targets is minimal (since it won't 
be built unless asked for), and the maintenance burden becomes easier -- as 
development continues keeping a static "in review" version becomes more 
costly.

It would also allow work to begin in earnest on setting up public testing of 
the backend, integrated into the usual buildbot structure.

I realise it's a massive set of changes and it needs to be subjected to 
further review, but back when experimental targets were being discussed there 
was talk of reviews carrying on during that phase. Is that still a viable 
plan? Or can we construct some kind of roadmap for clarity?

Cheers.

Tim.