Blacklist based on window titles

[Daxtorim]

This allows users to stop windows from getting tiled based on the window title (client.caption).

Once again, no clue why QT Designer completely rewrites config.ui

[zeroxoneafour]

Would it be possible to add an invert feature to the titles setting to whitelist windows by title?

[zeroxoneafour]

Never mind, I don't really know how exclusion would work with two whitelists. I may add support for this later. Code looks good, thank you!

[Daxtorim]

I don't really know how exclusion would work with two whitelists.

Yeah, I struggled with that one as well for a while which is why I didn't include anything here. But we may have been overthinking this.

Literally just

--- a/src/util.ts
+++ b/src/util.ts
@@ -62,7 +62,7 @@ export function doTileClient(client: KWin.AbstractClient): boolean {
     for (const i of config.filterClientCaption) {
         if (i !== "" && cap.includes(i)) {
             filterProcessCache.add(cap);
-            return false
+            return config.useCaptionWhitelist;
         }
     }

If the filter is configured with an explicit window title we act accordingly when encountering it, but take no action when any other title is checked. So when you are whitelisting "Title-1" and you encounter "Title-a" you don't block "Title-a", you just ignore it there and go on to check for process name. And if the process name is not blacklisted the window gets tiled despite the title not matching.

So instead of the title filter being a "yes/no" question it becomes a "yes/maybe" question and in case of a maybe we let the process filter decide. You get two options "Never tile" and "Always tile" based on window title and any title not configured for either options just gets passed on to the process filter to make a call.

Does that make any sense? I'm gonna see if I can make something up that better conveys what I mean.

[zeroxoneafour]

Yeah, the way they work together brings up a lot of confusion if they both have inverses. If one inverse is checked but not the other, which will overrule the other? Will the engine assume to tile or not to tile? A future, more simple solution would be to put a single global inverse, but it's probably fine for now.