[PATCH] D116821: [DebugInfo][InstrRef] Move instr-ref controlling flag out of TargetOptions

David Blaikie via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Sat Jan 22 12:17:59 PST 2022


dblaikie added a comment.

In D116821#3263746 <https://reviews.llvm.org/D116821#3263746>, @xbolva00 wrote:

>>> it's probably reasonable to revert, analyze, and see if there are some extreme cases
>
> Revert… So (silent) users, which are now or would be (soon new release) happy with better debug info, would just lose this feature?

Similarly, they could use the flag to continue to opt in to the behavior.

> Otherwise I dont know why many new bigger things in llvm go under own flag if not for situations like this one.

Generally if it ends up being a feature where a small subset of users have an esoteric need compared to the norm, they could continue to use the flag to opt out.

But if the tradeoff is more broadly unfavorable, then that probably indicates the default should reflect that.

It's unclear where we are in this balance without numbers in either direction.

But also if the observed regressions are larger than the documented/intended, that's probably reflective of some considerations that hadn't been accounted before before committing, and so worth reverting to known good and investigating those considerations to make sure they're understood before moving forward.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D116821/new/

https://reviews.llvm.org/D116821



More information about the llvm-commits mailing list