[PATCH] D139965: [RISCV] Add a bit to TSFlags to mark SignExtendingOpW instructions for SExtWRemoval.

[Alex Bradbury via Phabricator via llvm-commits]

asb added a comment.

In D139965#3995784 <https://reviews.llvm.org/D139965#3995784>, @reames wrote:

> Just curious, what was the motivation here?  The resulting code appears much less clear to me.
>
> If this was a compile time issue, did you consider using e.g. a denseset membership test instead?

>From my perspective it's really a style / taste issue (and to be honest when reviewing the patch I hadn't considered much that some might prefer the old way). Is it clearer to mark these instruction properties in tablegen where the instructions are defined, or in a separate helper function. In general I like the tablegen property approach, particularly if it's plausible the flag might be queried from multiple places in the future. But I don't feel overly strongly on this.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D139965/new/

https://reviews.llvm.org/D139965