frame opacity for focused or !focused, and shadow opacity that follows

[ortango]

for awhile i've used a version of the following patch:

diff file

diff -ruN a/src/picom.c b/src/picom.c
--- a/src/picom.c	2024-04-19 20:03:13.019651286 -0400
+++ b/src/picom.c	2024-04-19 20:05:57.516322479 -0400
@@ -926,7 +926,7 @@
 			*fade_running = true;
 		}
 
-		if (win_has_frame(w)) {
+		if (win_has_frame(w) && !w->focused) {
 			w->frame_opacity = ps->o.frame_opacity;
 		} else {
 			w->frame_opacity = 1.0;
@@ -1024,7 +1024,7 @@
 		log_verbose("Window %#010x (%s) will be painted", w->base.id, w->name);
 
 		// Calculate shadow opacity
-		w->shadow_opacity = ps->o.shadow_opacity * window_opacity * ps->o.frame_opacity;
+		w->shadow_opacity = ps->o.shadow_opacity * window_opacity * w->frame_opacity;
 
 		// Generate ignore region for painting to reduce GPU load
 		if (!w->reg_ignore) {
diff -ruN a/src/renderer/command_builder.c b/src/renderer/command_builder.c
--- a/src/renderer/command_builder.c	2024-04-19 20:03:13.019651286 -0400
+++ b/src/renderer/command_builder.c	2024-04-19 20:50:31.139727994 -0400
@@ -160,7 +160,7 @@
 	}
 	log_region(TRACE, &cmd->mask.region);
 	cmd->blit = (struct backend_blit_args){
-	    .opacity = layer->opacity,
+	    .opacity = w->frame_opacity,
 	    .max_brightness = 1,
 	    .mask = &cmd->mask,
 	    .ewidth = layer->shadow_size.width,

it sets the frame opacity as the configured frame opacity only for unfocused windows, and 100% for the focused or force focused windows.
also uses the window's frame opacity to determine the shadow's opacity.

it would be a nice feature to have - and the new backend/renderer together seem to calculate damage correctly without any extra work (? - a few commits ago i needed to use a inactive-opacity and fading to workaround this).