> On 1 Sep 2026, at 09:08, Jeff Law <[email protected]> wrote: > > > > On 9/1/26 1:07 AM, Richard Biener wrote: >> On Tue, Sep 1, 2026 at 7:53 AM <[email protected]> wrote: >>> From: Kyrylo Tkachov <[email protected]> >>> >>> The unsigned_integer_narrow_clip pattern assumes that its comparison >>> is unsigned and its right shift is signed. nop_convert? does not >>> require an unsigned conversion. The :s suffix marks a single-use >>> value rather than a signed type. >> This sounds like LLM produced, nobody would think :s marks a signed type. > Yea, I had the same though moments ago. >
Sorry, I did use an LLM to flesh out (just) the commit message in my hurry to close out the regression and I didn’t review it well enough for such nonsense. I’ll rephrase this myself before committing. Kyrill > Jeff
