Number of problems found in Windows 2.3 build (21212)

[Kenner]

I've noticed a few problems when playing with Destructavator's latest Windows build of 2.3 (rev. 21212), some of which have been around for awhile, so I thought it best to report them.

1. When I start a new game, as soon as I click on the second message in the Maillist (Proposal:Electromagnetic Rifle), the game crashes to the desktop with an assertion error:

File: F:\UFOai\src\renderer\r_state.c
Line 58
Expression: texnum > 0

Happens every time. If I run the game in safe-mode, the error doesn't occur (but of course sound is gone).

2. A couple of terror missions in South America (specifically, city name is Cayenne), the text reports an attack on an "African" village.

3. I have a jungle-based terro mission that won't run. Crashes to desktop with following assertion:

File: F:\UFOai\src\client\cl_le.c
Line: 117
Expression: lm -> model

4. I know stairs and pathfinding are apparently buggy at the moment. However, there are many areas on all maps that I cannot move onto at all. It's almost like there are invisible lines around some tiles that soldiers can't cross. Is this the problem with pathfinding?

Thanks,
Ken

[Destructavator]

Regarding point #4, I've seen this too, and I suspect it is because up through that version the map compiler (ufo2map.exe) was worked on and tweaked many times and yet most of the maps were not rebuilt because the actual map source files were unchanged, and the script for compiling them was set at the time to only rebuild updated map source files.

I haven't tried this with an installed development build, but I'd imagine it would be possible to go to the folder "contrib" and then "scripts" under the installed folder and run the "compile_maps" batch file with the "/clean" switch after it to rebuild all the maps with the updated compiler, and this could probably be done without Codeblocks, without SVN, or any other programming tools, and be very easy to do.

I must warn you - rebuilding all the maps can take hours, it might be best to do it overnight if you try this.

For the next build I upload and link to, I plan to do this beforehand so this won't be necessary.

I'll try to upload a new one with re-compiled maps within a week, if my schedule permits.


Regarding your other points, the developers will need to address those (if they aren't on vacation for the holidays).

[BTAxis]

Rebuilding maps is a lot faster if you pass -quant 6 as a parameter. The shadows will look like crap though.

[DuKe2112]

I already askeed last time and didn't get a response WHERE do you pass the parameter -quant 6, to which command?
The only thing that comes to mind is 'make maps', but that's doesn't know quant.

[BTAxis]

It's a parameter to ufo2map. There is probably a "fast" target in your makefile that passes the parameter.

[DuKe2112]

Ah, thanks.

[Duke]

It's almost like there are invisible lines around some tiles that soldiers can't cross.

Very interesting. Do you mean 'tile' like in
- one soldier's step per tile or
- the building blocks that random maps are made of ?

rev 21212 already contains one fix to ufo2map. Unfortunately, it fixes only some of the problems.
So yes, recompiling the map could help. If the problem persists, please report it (preferably to the bug tracker).

[geever]

I recompiled all my maps recently I think just after that change and there are some RMA problems reintroduced like RMA tile accessible only from one direction. (tested on +farm)

-geever

[Kildor]

If it may help.
there are some problems with brushes with nodraw flags on some faces. But it doesn`t have 100% correspondence. Look to debug_map 1

[DuKe2112]

Duke i think he ment steptiles, at least I have that problem too.
But the strange thing is, so far I have only encountered those between buildigs, meaning along the block tile edges, but not the whole edge only one or two steptiles wide.
Forgot to make a screen though.

[Duke]

Ok, we have (at least) 3 different types of walking-bugs here:
1. nodraw
2. compression
3. RMA tile border

1. That one is new to me. Thx for the hint, Kildor. Where can I find more info on this ? Tracker ? Forums ? ...?

2. the compression bug had the potential to corrupt 'single' bytes of the routing table. several hundred per map, but most of them were not critical to walking. It's highly unlikely that the corruptions would produce an 'invisible wall' longer that 2 or maybe 3 steps.
However, it's fixed. So by recompiling the map you can exclude this bug.

3. I'm also suspecting a bug in RMA. I would expect it to produce impassable tiles at or very close to the border/corners of RMA-tiles. So if you encounter a group of tiles that are impassable for no apparent reason (on an already recompiled RMA map), a screenshot would surely help. It should
- have medium zoom (so we can identify the RMA-tiles involved) and
- mark *all* the impassable tiles (so we can recognize the pattern).

[canahari]

3. I'm also suspecting a bug in RMA. I would expect it to produce impassable tiles at or very close to the border/corners of RMA-tiles. So if you encounter a group of tiles that are impassable for no apparent reason (on an already recompiled RMA map), a screenshot would surely help. It should

- have medium zoom (so we can identify the RMA-tiles involved) and

- mark *all* the impassable tiles (so we can recognize the pattern).

Hallo!

I created some screenshots about this problem. (Well, if there is really just one.) This was on an Athens farm map, today, 3-days-old SVN source, Linux version, OpenSUSE 11.1.

impassable.jpg: This is the area on which I first encountered the problem. The soldiers can't enter the brown field, not even crouching of course.

impassable-area.jpg: I kept trying,  and I found the followings: for the soldier marked with the red arrow, the whole area seems impassable. However, I had a soldier near the right corner, and he found a way through the upper gate. (Yellow arrows.) That is passable, and if he goes through that then all the brown area seems passable. I marked the impassable tiles with blue, passable are red.

area.jpg: Much more weird - I found an impassable area in the open field (between the buildings) to be seen on this image. 




[attachment deleted by admin]

[canahari]

Eh, I cannot upload any more images
They can be seen here: http://www.flickr.com/photos/canahari-debug/

area-marked.jpg: Here I marked the tiles passable to the soldier.

area2-marked.jpg: This is the same area, between two buildings (the dead green alien is the same one as on the last pic). Red tiles are passable, blues aren't. On the right that is the same brown field here that I mentioned above. The gate that can be seen here is too impassable, so it seems there are only one passable gate to this field.

[Kenner]

impassable-area.jpg: I kept trying,  and I found the followings: for the soldier marked with the red arrow, the whole area seems impassable. However, I had a soldier near the right corner, and he found a way through the upper gate. (Yellow arrows.) That is passable, and if he goes through that then all the brown area seems passable. I marked the impassable tiles with blue, passable are red.

area.jpg: Much more weird - I found an impassable area in the open field (between the buildings) to be seen on this image. 

Canahari,

Thanks for posting these screenshots. These are exactly the problems I was talking about, especially the impassable areas in open fields. If you look closely enough, you can almost see lines on the screens where the two blocks of areas are joined together and it seems like these joints are acting as impassable walls.

Ken

[Duke]

Thx canahari, good work

I've been able to narrow it down to a problem that's already in the tile *before* assembly. Investigating...

Currently no more screenshots needed.