Client rewrite β-testing

[shubla]

Trying to download test client, but .jar files download reaally slowly!

[loftar]

As another small status update, the SlotRemoved bug, the "indefinitely waiting for mapdata" bug, the Phong-shading null-pointer bug and a couple of unreported deadlock bugs should be fixed now.

[jordancoles]

As another small status update, the SlotRemoved bug, the "indefinitely waiting for mapdata" bug, the Phong-shading null-pointer bug and a couple of unreported deadlock bugs should be fixed now.

Nice

[MagicManICT]

As another small status update, the SlotRemoved bug, the "indefinitely waiting for mapdata" bug, the Phong-shading null-pointer bug and a couple of unreported deadlock bugs should be fixed now.

Good. Getting stuck waiting for map to load and log out and in every time was driving me nuts.

[shubla]

Code: Select all

java.lang.Error: Unexpected non-local exit
   at haven.render.RenderTree$TreeSlot.remove(RenderTree.java:488)
   at haven.render.RenderTree$TreeSlot.add(RenderTree.java:455)
   at haven.render.RenderTree$TreeSlot.add(RenderTree.java:356)
   at haven.render.RenderTree$Slot.add(RenderTree.java:343)
   at haven.AudioSprite$Ambience.added(AudioSprite.java:160)
   at haven.render.RenderTree$TreeSlot.add(RenderTree.java:452)
   at haven.render.RenderTree$TreeSlot.add(RenderTree.java:356)
   at haven.render.RenderTree$Slot.add(RenderTree.java:343)
   at haven.ResDrawable.added(ResDrawable.java:58)
   at haven.render.RenderTree$TreeSlot.add(RenderTree.java:452)
   at haven.render.RenderTree$TreeSlot.add(RenderTree.java:356)
   at haven.render.RenderTree$Slot.add(RenderTree.java:343)
   at haven.Gob.added(Gob.java:520)
   at haven.render.RenderTree$TreeSlot.add(RenderTree.java:452)
   at haven.render.RenderTree$TreeSlot.add(RenderTree.java:356)
   at haven.render.RenderTree$Slot.add(RenderTree.java:343)
   at haven.Gob$Placed.added(Gob.java:757)
   at haven.render.RenderTree$TreeSlot.add(RenderTree.java:452)
   at haven.render.RenderTree$TreeSlot.add(RenderTree.java:356)
   at haven.render.RenderTree$Slot.add(RenderTree.java:343)
   at haven.MCache$Grid$Flavobjs$1.added(MCache.java:175)
   at haven.render.RenderTree$TreeSlot.add(RenderTree.java:452)
   at haven.render.RenderTree$TreeSlot.add(RenderTree.java:356)
   at haven.render.RenderTree$Slot.add(RenderTree.java:343)
   at haven.MCache$Grid$Flavobjs.added(MCache.java:188)
   at haven.render.RenderTree$TreeSlot.add(RenderTree.java:452)
   at haven.render.RenderTree$TreeSlot.add(RenderTree.java:356)
   at haven.MapView$MapRaster$Grid.tick(MapView.java:596)
   at haven.MapView$Terrain.tick(MapView.java:657)
   at haven.MapView.tick(MapView.java:1440)
   at haven.Widget.tick(Widget.java:702)
   at haven.GameUI.tick(GameUI.java:856)
   at haven.Widget.tick(Widget.java:702)
   at haven.UI.tick(UI.java:145)
   at haven.JOGLPanel.run(JOGLPanel.java:362)
   at java.lang.Thread.run(Thread.java:748)
   Suppressed: #<Resource sfx/ch-music>
      at haven.Resource$Pool$Queued.get(Resource.java:383)
      at haven.Resource$Pool$Queued.get(Resource.java:355)
      at haven.ClipAmbiance.added(ClipAmbiance.java:246)
      at haven.render.RenderTree$TreeSlot.add(RenderTree.java:452)
      ... 34 more
Caused by: java.lang.NullPointerException
   at haven.ClipAmbiance$Glob.hashCode(ClipAmbiance.java:66)
   at java.util.HashMap.hash(HashMap.java:339)
   at java.util.HashMap.get(HashMap.java:557)
   at haven.ActAudio.intern(ActAudio.java:267)
   at haven.ClipAmbiance.removed(ClipAmbiance.java:254)
   at haven.render.RenderTree$TreeSlot.remove(RenderTree.java:483)
   ... 35 more


Logged in and got some lore event or such, UI is partially loaded, but otherwise everything is black. The lore/something window only has title that says "Hey, listen!" and is otherwise empty.

[fallout]

Code: Select all

java.lang.RuntimeException: Delayed error in resource gfx/tiles/swamp/flavobjs/bladegrass (v2), from local res source
   at haven.Resource$Pool$Queued.get(Resource.java:386)
   at haven.Resource$Pool$Queued.get(Resource.java:355)
   at haven.ResDrawable.<init>(ResDrawable.java:42)
   at haven.MCache$Grid.makeflavor(MCache.java:222)
   at haven.MCache$Grid.getfo(MCache.java:234)
   at haven.MCache.getfo(MCache.java:535)
   at haven.MapView$Terrain$2.getcut(MapView.java:646)
   at haven.MapView$Terrain$2.getcut(MapView.java:644)
   at haven.MapView$MapRaster$Grid.tick(MapView.java:577)
   at haven.MapView$Terrain.tick(MapView.java:657)
   at haven.MapView.tick(MapView.java:1440)
   at haven.Widget.tick(Widget.java:702)
   at haven.GameUI.tick(GameUI.java:856)
   at haven.Widget.tick(Widget.java:702)
   at haven.UI.tick(UI.java:145)
   at haven.JOGLPanel.run(JOGLPanel.java:362)
   at java.lang.Thread.run(Unknown Source)
Caused by: haven.Resource$LoadException: Load error in resource gfx/tiles/swamp/flavobjs/bladegrass(v2), from local res source
   at haven.Resource$Pool.handle(Resource.java:447)
   at haven.Resource$Pool.access$1100(Resource.java:332)
   at haven.Resource$Pool$Loader.run(Resource.java:583)
   ... 1 more
Caused by: java.lang.NullPointerException
   at haven.FastMesh$MeshRes.init(FastMesh.java:262)
   at haven.Resource.load(Resource.java:1402)
   at haven.Resource.access$600(Resource.java:40)
   at haven.Resource$Pool.handle(Resource.java:435)
   ... 3 more


Found a nice chunk of my village I cant visit yet with this client.

[loftar]

As another status update, I do believe I've fixed the vast majority of the outright crash bugs by now, including at least, I believe, all the crash bugs mentioned thus far in this thread, so it would be nice if y'all could test it a bit more and see if you discover any new ones. I've also fixed most of the graphical glitches mentioned in the thread, but there are still a few remaining, I'll update the OP to contain the currently known ones.

[loftar]

the only down side is that i think it didnt do much for the animation lag when there are a lot of players on screen as like 40/50 animals still drops my fps to 30-

I'm curious if there are special circumstances or something that cause that, because I can have over 100 rendered player objects on screen before my FPS starts dropping significantly below 60 (with :par on):

ss.png (696.29 KiB) Viewed 884 times

I do expect the implementation of GPU-side skinning to also make an impact on this, but I'm still curious. Is continuous movement movement/changing of poses for the involved players important to cause this? Does it happen mostly while also clicking incessantly? Any other factors? Of course, the objects I test this with aren't "real" players, but I can't imagine what different the client would see between these test objects and real players except them not moving around and switching poses. Perhaps they use fewer clothing layers than realistic players, if anything. I'm really quite curious.

For the record, the relevant parts of the hardware that I test on include a Core i5 2400, and a GTX 750.

[jordancoles]

That many characters clicking around on screen would likely add more fps loss. I noticed a drop in fps when boating with large groups of people, ie. 4 full rowboats all traveling down the same river together

Also, all animals. Spawn in 100 cows and pigs instead of people and have them derp around on screen

[loftar]

I noticed a drop in fps when boating with large groups of people, ie. 4 full rowboats all traveling down the same river together

Was that something you noticed on the stable or on the new client?

Also, all animals. Spawn in 100 cows and pigs instead of people and have them derp around on screen

200 cows, 100 pigs:

ss.png (629.51 KiB) Viewed 856 times

Also, as another note, a tip on performance is to use the G1 garbage collector, but I'm not sure in exactly what version of Java it was made the default. It could be Java 9, but it's definitely the default in Java 11. It does however not exist at all in Java 8, which I believe is what most people still use. The reason why I haven't recommended it previously is because it makes allocations ~30% slower in my testing, meaning it's not a great match for the current (stable) client, but the new client doesn't have any problems with that, since its allocation-heavy threads generally aren't the bottleneck. The advantage of the G1 GC is that it heavily reduces the need to "stop the world" for garbage collection, which is a significant source of major stutters.