[supergoofy]

Another crash in r1467 and dukeplus 2.03 + new dpcons fix


WARNING: SE23 i<0!
Fatal Signal caught: SIGSEGV. Bailing out.
1670 319048160
53859: 31337 {
53860: 8272 state 319047244
53862: 8273 ifrnd 48 319048140
53865: 8274 spawn 2270
53867: 8275 debris 2408 12
53870: 8276 killit
53871: 31337 }
53872: 8278 else 319048176
53874: 8279 action 319047768
53876: 31337 }
53877: 8281 enda
53878: 8284 ifrnd 64 319048220
53881: 31337 {
53882: 8286 ifnosounds 319048216
53884: 8287 soundonce 336
53886: 31337 }
53887: 8289 else 319048264
53889: 8290 ifrnd 64 319048264
53892: 31337 {
53893: 8292 ifnosounds 319048260
53895: 8293 soundonce 337current actor: 162 (1062)
g_errorLineNum: 23912, g_tw: 124



I used shrinker and shoot few times and then freezer and shoot few times, then boom SIGSEGV

[Stabs]

hmm got in and tried it dp203 unfixed cons

i fired of hundreads of shrinker and freezer shots without a crash, shrunk that blue medic and froze him a number of times, but no crash, the shrink problem was there but it caused no crash

the only problem i had was with starting eduke32 with dp, but deleting the cache files fixed that.

[supergoofy]

the crash is only related to the use of freezer


load E1L1

shoot with the freezer at the location that you see in the picture




SIGSEGV crash


btw, the new dpcons fixed the bug with shrinking the mediks aliens

[Stabs]

tried that too, i cannot seem to replicate this crash

[supergoofy]

I found what is the problem:

First it's eduke32 fault

I run the game using long path and including in the patch the "." , "-" , "_" characters

that's why it crashed.

When I changed name to the folders (I used short names). it worked fine.

this is an on-disk textured cache bug that appears when you have long path and/or characters like "." , "-" , "_" (I'm not sure if one of the characters is triggering the bug or it is only triggered by the long path). like I had

it is strange though that happens with the freezer.

either way now with shorter path all work ok:
C:\Games\Duke.Nukem.3D\High.Resolution\

[TerminX]

Well, post the directory names as well as the command lines you were using and I'll take a look into it. It must be related to the long directory names (how long are we talking here?) and not the characters, because I just tried Duke with some of the most hilariously badly named directories I've ever seen in terms of the characters you mentioned and it was fine.

[supergoofy]

The crash happens when:

all files and folders are inside the folder:

C:\Games\Duke.Nukem.3D.Atomic.Edition.with.High.Resolution.Pack-[Test_15. 07.2009]\dn3d_hrp-07-07-2009\


In autoload folder I have:
duke3d_hrp.zip
hrp_update.zip
maphacks.zip
xxx_lite.zip

DUKEPLUS_RESOURCES.zip is unpacked. Thus I have the folder DukePlus\DUKEPLUS_RESOURCES\

I play with:
eduke32 -geduke32_mus.zip -jDukePlus -hdukeplus.def -xeduke.con


Because I don't use the -game_dir command, that's why I have unpacked DUKEPLUS_RESOURCES.zip

[Tea Monster]

Did you try simplifying the name? Specifically, remove any instances of "-" out of the name of the files or folders. "_" should be fine and "." but remove the hyphen from "Pack-[Test..." as it may be thinking that the hyphen is a command line switch.

[supergoofy]

This was a path example for TX so that he can find the error.

I already use a simpler and smaller path to avoid crashes.

[Master Fibbles]

Enemies fly when shrunk. That isn't new and it isn't a bug as far as I know. I remember the Octobrain flying around after you shrunk it. I am also pretty sure they still shot too, but I'd have to go play the game again.

[supergoofy]

the bug was that they stayed shrunk and shot at you, fly etc. etc.. and they stayed shrunk forever.

the thing is that in original game when you shrunk them they cannot harm you. but you can kill them any way you want, e.g. shoot a rocket at the tinny/shrunk alien and he will die/vanish

in dukeplus there is a bug when advanced ai is on. when you shrunk them you can only kill them by squashing them with the foot. e.g. if you shot a rocket at a shrunk enemy, he is invulnerable and don't die. they only way is by squashing them with the foot.

[Fox]

The original game and console ports already used to have that bug. If an enemy (with exception of Lizman and Newbeast) shows its flintching pose while shurunk, he will start to act as a normal enemy while shrunk.

[Danukem]

supergoofy, on Jul 15 2009, 06:41 AM, said:

in dukeplus there is a bug when advanced ai is on. when you shrunk them you can only kill them by squashing them with the foot. e.g. if you shot a rocket at a shrunk enemy, he is invulnerable and don't die. they only way is by squashing them with the foot.

I'll fix that, but would you please stop posting about Duke Plus bugs in the Polymer thread?

[BlackDays]

Can I get an explanation as to what specpower and specfactor do? I am using specular maps and I can't really tell the difference on the textures I'm using (wet caverns and damp stone bricks.)

Also, parallaxbias and parallaxscale. I understand that you have to use a height map, but is there a way to get a decent parallax effect while still using a normal map for the shaders? This tends to ruin the per pixel lighting on my textures. Also, it seems to ignore my specular maps when I use parallax effects.

[Devon]

Well this surely is a great year for Duke nukem 3D.


Sweet work plagman! cant wait to get my hands on that and start mod again.

i got some plans on a survivor horror themed game and now it seems Duke nukem 3D will be my choice after seeing this new renderer.

Thank you!

[supergoofy]

the problem with the freezer and long path was due to gcc 4.4.0

gcc 4.4.0 broke many things

that's why I insisted to keep 4.3.3

KEEP AWAY FROM GCC 4.4.0

[KillerBudgie]

supergoofy, on Jul 16 2009, 10:21 AM, said:

KEEP AWAY FROM GCC 4.4.0

Why? Too many bugs?

[supergoofy]

for example:

Yesterday I compiled r1461 with gcc 4.4.0 and though I didn't have any errors during source compilation, the compiled exe has some problems and give SIGSEGV crashes.

Today I compiled r1461 with gcc 4.3.3 and these problems are gone.

I'm asking sorry for troubling TX about these errors.


GCC 4.4.0 is not compatible with eduke32. And eduke32 is not the only program that has problems with gcc 4.4.0

That was the reason why I kept using gcc 4.3.3 for so long. Because gcc 4.3.3 is proven trustworthy and reliable.

[TerminX]

Yeah. I'm guessing 4.4.1 will be a lot better.

[Dr. Kylstein]

Are the light sector-effectors still working for everybody else? My map went dark! (except for the switches, which are still too bright in my opinion.) I can't remember which revision they were working in, but I still have a copy of that source.

Edit: Now the switches don't light either, even though nothing changed. The lights still work in E1L1 though.

[The Commander]

They did get broken for a time but they are fixed with the latest revisions. Which one are you using?

[Dr. Kylstein]

The Commander, on Jul 16 2009, 04:13 PM, said:

They did get broken for a time but they are fixed with the latest revisions. Which one are you using?

This is in r1461.

[supergoofy]

are you using gcc 4.3.3 to compiled the source?

[Jblade]

Is there any way to avoid or stop the 'random ID is the same' error for multiplayer? it normally happens alot when we have more than 2 players and it's a huge pain for us.

[atlas]

any chance we can get some more progress shots? bug screenshots don't do much for me i'm afraid.

[Dr. Kylstein]

supergoofy, on Jul 16 2009, 04:21 PM, said:

are you using gcc 4.3.3 to compiled the source?

"4.3.1-ubuntu2" I've never had any problems with it before.

[d3drocks]

is this ever going to be released?

[Danukem]

Do all of the mdflags work in Polymer yet? Last time I checked they didn't. Also, what about tints and the tint flag for making a model/texture greyscale? (that didn't work last time I checked either).

[Muelsa]

something is wrong with
setactor [THISACTOR]. angoff gamevar
Angoff value is greater with polymer.
the Model Rotate too much compared to the Angoff's value.

[Danukem]

Muelsa, on Jul 17 2009, 10:54 AM, said:

something is wrong with
setactor [THISACTOR]. angoff gamevar
Angoff value is greater with polymer.
the Model Rotate too much compared to the Angoff's value.

I think I remember Plagman saying that model offsets were done incorrectly in Polymost and he corrected them in Polymer. I'm not sure if that has anything to do with this though.