On Mon, Aug 05, 2024 at 12:32:54AM +1000, Mark Jamsek wrote:
> The below diff is minimally invasive and very simple. We ensure a proper
> ed script is generated in the case of diffs involving a malformed line
> but when writing the merged file, we elide the added terminating newline.
> 
> Regress is happy with the fix, which appears to address naddy's simple
> repro, but I'd like to confirm that it also addresses the llvm case.
> I want to add a regress of naddy's repro, too, but have run out of time
> tonight.

Looks good to me, ok by me if naddy can confirm the issue is fixed.
I agree tests would be good to have of course.