Hi Eli: Please give a look to the attached patch I am still testing it. But just to check if it solves all the issues you see. Best Ergus On Tue, Dec 17, 2019 at 05:22:28AM +0200, Eli Zaretskii wrote: >> Date: Mon, 16 Dec 2019 21:31:56 +0100 >> From: Ergus >> Cc: emacs-devel@gnu.org >> >> I had build a wrong branch where I have actually added >> >> NILP (face->lface[LFACE_EXTEND_INDEX]) >> >> to the condition you mention. >> >> Actually the question is why there is only: >> >> face->box == FACE_NO_BOX >> >> Maybe we have to remove it and rely only in the extend attribute? > >If the extend attribute is set, but the background color is the same >as the default, and there's no box/underline/overline/strikethrough >attributes set, then we can still return early, no? >