EE Development Release 87.8193.35-36

Announced here.

Detail here.

Can’t comment yet as the Beamdog Client version is “to follow”.

Am told Beamdog Client is updating now as well.

1 Like

1.33 GB Download on BD Client. Must be a lot of people D’Ling it as it is a slow download (600k-800k ass opposed to 3meg a second).

Also announced on BD Facebook page.

TR

1 Like

Lots of goodies and some fun 2da merging…

1 Like

Nothing to bad, some are having trouble with the custom damage. Jasperre has a break down here. LINK

Wow…!!
Tremendous news !
Even more fabulous buttons, dials, lights, knobs, & switches… :cookiemonster:
And I hadn’t even finished migrating into .34 yet…

[sighs]…

I suppose I will have to remain at the back of the “gog” bus for a good while longer…

“Here I must raise my ebenezer ; hither by thy help I’m come…”
Thank you, developers and curators for all your fine works over many years…

Long life to Nwn !
:nwnee:

From the Facebook message -

A new patch for NWN:EE is being worked on, and you all are cordially invited to participate in playtesting—or just playing—using the new Development branch on Steam/GOG Galaxy with bugfixes & new features.

Seems like [ P&SC ] Pentagrams & Summoning Circles v1.45 : real-time 3D geometry works properly again. In fact it seems (could be wrong here) as though all the beam vfx run faster now.

New external script editor (tried with notepad++) works OK. Which is great, although it does seem to mean that debugging now has to be done via the build menu. Would be nice to have a check box to select between the built-in editor and the external editor.

Haven’t checked on the fixed waypoint face east bug yet but that can wait until tomorrow now.

TR

1 Like

BD Client download now fast.

As we might reasonably expect, saves under 8193.35-36 don’t work with the Retail release unless you tweak the version in the module.ifo.

The waypoint/post issue appears to be fixed. No sign of it in @Axe’s Killer Walkway Points system demo module.

TR

1 Like

I see no mention of pathfinding there. I have no idea what the new code for pathfinding was supposed to solve. But even with .34 which brought some fixes to this new crap, the patfinding is still worse than ever and my summons are getting stuck by the regular tileset walls very often.

Was that finally fixed in .35 ?

Also funny how many issues is now under “fixed” list that was before not a problem, not a bug or my own fault for doing things in a way they never were supposed to when I was reporting them to BD over the years…

For example: “Fixed a crash when unloading iprp_costtable.2da if it had blank lines or missing 2das.”
(This started plaguing my CPP ever since EE was released as the server might reference via this file 2das that client didn\t have.)

Also, no fix for familiars regaining AI control during possession and act on their own (thats new in EE)?

Dear @Shadooow (and others sharing these concerns),
pathfinding, like the RNG, is notorious in making humans see patterns where there are none. For 15 years there have been reports how the RNG in NWN is broken, yet all statistical analysis shows it to be perfectly valid. Doesn’t stop the player who rolled a 4 five times in a row to make assumptions.

Similarly, there have been numerous reports that pathfinding in EE has gotten worse, but no one has ever bothered to point out an exact case, always opting for vague statements such as yours. We have tests for pathfinding that we run on latest EE and on 1.69, and EE wins by a small margin in the quality of the path found. EE also wins by a large margin in the time it took to find such a path. These are the things we can actually measure and can state objectively.

Now, I’m not denying that there might be cases where 1.69 ends up with a better path. I’m just saying that no one has ever created a reliable repro of such a case for us to investigate. And it certainly should have been easy - Simply send us a save with an item on the ground at the destination, in a scenario where clicking to pick up the item gives a shorter path in 1.69 compared to EE. Anyone playing any module seeing these regressions could just drop any item on the ground, move back, save the game and send it. Yet no one has done it over the last 5 years. Doesn’t stop vague complaints though.

Yeah, I just don’t see the EE pathfinding as broken, and I’ve tested it quite thoroughly all through EE development. Is it perfect? No, but I wouldn’t expect it to be. Are there pathnode issues in the base game tiles? Yes, some, and these are easily fixed by taking a Text editor to the SET file while playing with the Render Pathnode option toggled on. CPP fixed a lot, and FUBAR fixed more.

Creatures sometimes get stuck trying to trace a route, especially in areas where there are lots of placeables painted. Makes me wonder if it might be a PERSPACE issue (e.g., I’m too big to fit through the meandering pathway left by all these placeables). My testing has also shown that pathfinding always wants to go by the most direct-line route in preference to meandering around obstacles. This is probably why lots of placeables can, and sometimes do, mess with it.

I also find that, if pathfinding is going to mess up, it usually around an obscurring mesh such as jutting wall (the crypt tileset is notorious for me) - maybe something to do with the AI wanting LOS to pathfind to?

That said, the MINOR pathfinding issues I’ve encountered just aren’t worth reporting. They’re too few and far between to create a reliable repro. But as to EE pathfinding being worse than 1.69, that’s just pure malarky - at least in my experience.

OH, I also want to extend a GIANT thank-you to the VOLUNTEER development team that brought us this amazing patch. Your efforts are appreciated.

2 Likes

Well maybe if you were actually opening and trying our repro modules that you demand to be posted alongside bugreports, peoples would be more willing to do that ground work for you.

I am not sure if I posted the videos of the pathfinding issues anywhere or to anyone. You might be correct and I might have kept it for myself. IIRC before I upload them to youtube you made the hotfix 34 that was supposed to address it.

Still you are straight up lying there. There are several instances where the EE pathfinding broke old modules and prevented some placeables to be used/activated. This was publically pointed out at vault forum with savegame posted alongside. I was even trying to help there and verify the issue, it should still be archived on vault forum.

This bugreport is also related to pathfinding change in EE - Attacking through jail-doors · Issue #406 · Beamdog/nwn-issues · GitHub, noone bothered to even reply that it is intented feature. Version 34 with its fixes to pathfinding did not change this, it is still possible to do, but wasn’t in 1.69.

I also privately reported two exploits to you or @niv maybe? related to pathfinding, but to be fair the first one appeared before the pathfinding was changed (but still wasn’t fixed, in .34 at least). The second one appeared after and wasn’t fixed in .34.

Ok so the videos uploaded while I was writing this, here are links:
Neverwinter Nights Enhanced Edition v85 8193 32 6ba2702a 2021 11 28 10 47 20 - YouTube - I think this has been fixed in .34
Neverwinter Nights Enhanced Edition v85 8193 32 6ba2702a 2021 11 28 10 37 53 - YouTube - this is still happening in .34
Neverwinter Nights Enhanced Edition v85 8193 32 6ba2702a 2021 11 28 10 50 01 - YouTube - this was fixed in .34, however the problem with door remains, not in this exact area, but it happens often elsewhere, whenever the PC is coming to doors from side, behind or above
Neverwinter Nights Enhanced Edition v86 8193 34 1 b8d939f8 2023 03 17 13 36 39 - YouTube - this was captured under .34

So lot was fixed in .34 already, I am simply posting the videos to prove that this is not in my head as you are implying. Still, whenever other players or summons are involved the patfinding is worse than ever. Will make more videos if you mean this and you no longer ignore me and my bugreports.

That might be because you as a “roleplay” style of builder are working/playing/testing in areas with lots of placeables where pathfinding might have always been a problem. I wouldn’t know as I am an “action” player and builder and Iam playing/testing/building areas with little to no placeables. Hence I do not recall any issues with pathfinding in vanilla 1.69. The few that were there were related to bad .SET data as you pointed out and I fixed these in CPP so this was not an issue inside the pathfinding itself. The other few were related to sloped tilesets or tiles for which no suitable pathnode exists - case of Crypt tileset Stairs down tile group (or existed with the new pathnodes being added now in this development version). So for me, when I now see pathfinding issue in pretty much every gameplay session on my placeable-less PW I am not exaggerating when I claim it is worse than ever was.

Attacking through jail doors is probably related to the BlockSight parameter in genericdoors.2da and doortypes,2da. It makes sense that you should be able to attack through a door you can see through (curtains too for that matter).

If it is indeed 2DA related then builders can certainly disable it easily enough.

As to the “roleplay” style of builder - you’re way off friend. I go pretty light on the placeables myself. I was referring to what I noticed in testing theories that plopping down lots of placeables could negatively affect pathfinding.

Yes I suspect that the BlockSight to 1 will fix that. But then you get jaildoors that block vision which is just wrong.

Even if this is intended behavior, it doesn’t make much sense that you can attack with large weapons like scythe through jail doors, does it? Or bare hands. Likewise not all jail doors have even big “holes” where even the dagger could be put through.

Yeah, its going to be hard to isolate weapon types that can attack through doors unless they add a column to baseitems.2da and can code it into the engine without killing other stuff.

Time for a feature request for 36 :slight_smile:

EDIT - might need to be wary changing that BlockSight parameter to 0 though, since some modules such as the OC have people in cells you need to talk with. Changing that might screw that up.

R.e. vid 1. Whether “collide-ables” are blocking or not was changed: it’s ~10cm from walkmesh now, which is a big increase from 1.69 (at least I don’t think the candelabras are part of tile). Surprised the other npcs were able to path to you…they used to give up way before that. There are tilesets that are grabbier than before like crypt tiles mentioned, old forest tileset is as well.

Haven’t seen conversation on what it’s a good idea to change perspace to or where to adjust use nodes on placeables in one place yet. I know there are places in the OCs that are more difficult to access now. But none of that would fall under pathfinding either.

The jail doors thing sounds like a data change and not pathfinding code. At some point a lot of the doors’ walkmeshes changed and that ended up causing a ton of grief. Data changes are generally a bad idea and I don’t think they’ll be happening anymore, for better or worse (can’t revert now or we re-break everything that adapted already).

The videos you linked all show genuinely bad pathfinding behavior, there is no denying that. But they are still:

  1. Videos and not repro modules
  2. Unclear if it is a regression from 1.69, since the same things did happen there occasionally too.

Nobody is saying that pathfinding is (or ever was) good in NWN. I am just saying that there has been no concrete evidence provided that the pathfinding logic is worse in 8193.35 (or .34) than it is in 1.69. The best way to prove that regression is a savegame that can be opened in both EE and 1.69, with unambiguous instructions (e.g. click on item on the ground), which shows the different behavior. I guarantee that such a mod would be looked at. We’ve been asking for this every time these claims are made and so far no one has taken us up on the offer.

Note that it is perfectly expected that EE and 1.69 generate a different path of same length in some cases. Such as if you’re directly in front of a pillar and want to go around it, EE might go right and 1.69 might go left; the two are equivalent distances, and it’s not considered a bug.