Sauerbraten engine development

by Aardappel on 03/03/2004 05:18, 1571 messages, last message: 03/14/2008 18:53, 1351324 views, last view: 12/09/2021 06:27

This thread is for discussion of Sauerbraten coding issues / implementation ideas etc.

   Board Index   

#1435: Re: new feature: full screen shaders

by guest57 on 05/28/2006 14:58, refers to #1434

Yeap, that makes it work...
I'll leave your gang to come up with a workaround as I'm not an OpenGL person. Meanwhile, nice effect :-)

reply to this message

#1436: Re: new feature: full screen shaders

by absinth on 05/28/2006 16:34, refers to #1435

http://developer.apple.com/graphicsimaging/opengl/extensions.html#GL_EXT_texture_rectangle

reply to this message

#1437: Re: new feature: full screen shaders

by Aardappel_ on 05/28/2006 22:04, refers to #1436

Yeah I knew about the existence of that extension... just on my card it seems to work without. I will add it, thanks.

reply to this message

#1438: Re: new feature: full screen shaders

by Passa on 05/28/2006 23:32, refers to #1435

How did you get the latest build? I am soo desperate to try these shaders :)

reply to this message

#1439: Re: new feature: full screen shaders

by absinth on 05/28/2006 23:33, refers to #1438

>How did you get the latest build?
CVS?

reply to this message

#1440: Re: new feature: full screen shaders

by Passa on 05/29/2006 08:45, refers to #1439

Anonymouse CVS has been refreshed? How do I access it with TortoiseCVS? It keeps trying to login differently with a password..

reply to this message

#1441: Re: new feature: full screen shaders

by MeatROme on 05/29/2006 12:03, refers to #1440

http://sourceforge.net/cvs/?group_id=102911
and read the part about
"Anonymous CVS"

HTH

reply to this message

#1442: Re: new feature: full screen shaders

by Passa on 05/29/2006 13:52, refers to #1441

Still get asked for a password and if I leave it blank it fails.

reply to this message

#1443: Re: new feature: full screen shaders

by Passa on 05/29/2006 13:57, refers to #1442

No matter, got it via http, worked fine.. just need to know how to activate the shaders..

reply to this message

#1444: Re: new feature: full screen shaders

by MeatROme on 05/29/2006 14:02, refers to #1442

just did a current check-out myself, so works fine for me. dunno where your prob lies. any error msgs?
getting asked for a password is normal, they say so on the page BTW.
did you replace "modulename" (in the second call after the AUTH call [which is anonymous and therefor with blank password]) with "sauerbraten"?!?

Anyway you might want to read
http://sourceforge.net/docs/E04/

reply to this message

#1445: Re: new feature: full screen shaders

by Passa on 05/29/2006 14:05, refers to #1444

All I know is that I got the 1.49 revision.. released by aardappel apparently 8 hours ago. Also, all I downloaded was the .exe, nothing else.

reply to this message

#1446: Re: new feature: full screen shaders

by Passa on 05/29/2006 14:25, refers to #1445

Ok got it working, a few problems, the writing is all messed.. but it looks nice. Needs ALOT of tweaking though, it was killing my eyes, I don't think I could have played with it like that.

http://members.iinet.net.au/~passarello/bug2.jpg

Thats the writing problem bug

Here is a problem with the current release that I mentioned before, it still exists in the latest build (although this screenshot is of the current release)

http://members.iinet.net.au/~passarello/bug1.jpg

Last time Aardappel, you dismissed it because I had BETA drivers, well I installed the 84.21 drivers, and the bug was still there. My specs are P4 3.2Ghz, 2GB ram and a Leadtek 256MB GeForce 7800GT

reply to this message

#1447: Re: new feature: full screen shaders

by rpointon on 05/29/2006 14:55, refers to #1446

The writing bug is because you don't have the necessary image resource files for the text (the text style was changed over the last month so that it has a black outline). I assume you only checked out the executable.

reply to this message

#1448: Re: new feature: full screen shaders

by Aardappel_ on 05/29/2006 22:31, refers to #1446

yes, get a full checkout, not just individual files. The gun is weird... anyone else have this?

reply to this message

#1449: Re: new feature: full screen shaders

by Passa on 05/29/2006 22:48, refers to #1447

On the writing, I got the newerchars.png file, and it wasn't fixed.

reply to this message

#1450: Re: new feature: full screen shaders

by nieb on 05/29/2006 23:44, refers to #1448

Yes, I also have the hudgun in close walls problem. Didn't think much of it though. Couldn't you just make it render on top of everything else?

reply to this message

#1451: Re: new feature: full screen shaders

by guest57 on 05/29/2006 23:45, refers to #1448

Which gun do you mean?

I note that if one model/file doesn't load/missing then all following files don't load - deliberate?

reply to this message

#1452: Segfault fix for malformed maps

by xoltra on 05/30/2006 01:45

I noticed that the sauer_client may crash when it attempts to load a malformed map. I fond that this was caused by attempt to free the "worldroot" when the "garbage in map" fatal error is encountered. I fixed it by making sure "worldroot" was either valid or null. You can find sauer-bad-map-crash.diff at the end of this page:
http://home.austin.rr.com/selliott4/sauerbraten

I think it would be nice to get segfaults caused by bad data before resurrecting the /getmap /sendmap functionality because memory corruption problems have potential to be buffer overflow exploits. But I don't know if that is possible in this case.

Also, I think it would be nice if sauer_client would log an error to the console and then reload the previous map if an attempt to load a map failed (as opposed to exiting). I didn't attempt to do this in my patch, but I think loadchildren() and loadc() could be modified to return true/false indicating whether the load was successful. If a failure propagates up to load_world() it could call itself with the previous map. This appeals to me because otherwise once it becomes possible to sendmaps to the server it will be possible to send a map that causes everyone's client to exit.

Also, you may have discussed this, but wouldn't it simplify things if the top level cube related entity was a single single large cube that contained the entire world instead of "worldroot", which is an array of eight cubes? For example, you wouldn't need to have two calls to loadchildren() in the code. Also, freeocta() could be a "del rootcube", or whatever. But it probably doesn't matter much.

reply to this message

#1453: Re: new feature: full screen shaders

by Aardappel_ on 05/30/2006 04:28, refers to #1450

ok, we'll scale the hudguns down a bit, see if that fixes it for you all on the next release. Actually, can you put a command "mdlscale 50" or so in the md2.cfg of the relevant hudguns, and see if that fixes it? if not, try lowering the number, tell me the lowest number at which the bug goes away.

no idea on the chars... anyone else?

reply to this message

#1454: Re: Segfault fix for malformed maps

by Aardappel_ on 05/30/2006 04:31, refers to #1452

really, if a map is malformatted, something is seriously wrong (corrupted download?), and there is not a lot of point investing time in recovering. In all my time, I NEVER had this happen for either cube or sauer, and I bet I have loaded the most maps of anyone :)

Infact, if it happens, it shouldn't just be ignored by loading another map.

reply to this message

   Board Index