Hi! On 2019-12-03T12:56:49+0000, Andrew Stubbs <a...@codesourcery.com> wrote: > On 02/12/2019 14:19, Thomas Schwinge wrote: >> Generally, I'm in favor if you'd consider such a thing (that in principle >> is just a copy/adapt of the existing cases) as obvious to commit (even >> more so with your "amdgcn port" maintainer hat on), especially so given >> that this has been/is blocking you, as Tobias told me more than once. > > I probably will do for incremental tweaks, but I left some stuff in this > one to see if you were awake (that's my story and I'm sticking to it).
Haha! ;-P > Here's what I have committed (and I just realized I forgot the new > fangled reviewed-by tag, sorry). No worries. That's just an experiment anyway -- seems to work/have been picked up for glibc, but not so much for GCC. About the alphabetic sorting that I've mentioned: sometimes we do that, sometimes we sort per 'GOMP_DEVICE_*'/'acc_device_t'/whatever ordering, sometimes special cases go to the beginning/end, sometimes it's "in order of appearance", or seemingly random, or any combination of all these. So there isn't a general guideline to follow. However, to make reading the code more easy/enjoyable, what I like to see, is at least some kind of local consistency, instead of just always adding new stuff to the end. (Where the latter might indeed be the right thing to do in certain cases, given that 'GOMP_DEVICE_GCN' is the last one in the list.) Grüße Thomas
signature.asc
Description: PGP signature