Server Help

LVZ/LVL Questions :: RE: dynamic runtime LVZ options

JoWie — Tue, 06 Jan 2009 22:40:00 GMT

Author: JoWie
Posted: Tue Jan 06, 2009 5:40 pm (GMT -5)

You can easily detect bomb fires, but not hits.

Also: https://forums.minegoboom.com/viewtopic.php?t=7506

ASSS Questions :: RE: Loading pymod

Dr Brain — Tue, 06 Jan 2009 22:15:05 GMT

Author: Dr Brain
Posted: Tue Jan 06, 2009 5:15 pm (GMT -5)

I'm including the my preliminary changes to make asss run and log players in. I haven't exhaustively tested the system. I'd wager there are a few more bugs lurking around, and would appreciate any feedback. I've also attached a 64-bit security.so.

Clientset has some code that needs updating:

Code: Show/Hide

--- src/core/clientset.c 2ad6f3ce5cb2ba50fd737032e5ae23d1684a804c
+++ src/core/clientset.c 9a5e4bd79825c76d979f5131cbeea90fdf53be0d
@@ -389,22 +389,22 @@ local void do_mask(
   overridedata *od2)
{
int i;
- unsigned long *s = (unsigned long*)src;
- unsigned long *d = (unsigned long*)dest;
- unsigned long *o1 = (unsigned long*)od1->bits;
- unsigned long *m1 = (unsigned long*)od1->mask;
+ unsigned int *s = (unsigned int*)src;
+ unsigned int *d = (unsigned int*)dest;
+ unsigned int *o1 = (unsigned int*)od1->bits;
+ unsigned int *m1 = (unsigned int*)od1->mask;

if (od2)
{
-    unsigned long *o2 = (unsigned long*)od2->bits;
-    unsigned long *m2 = (unsigned long*)od2->mask;
+    unsigned int *o2 = (unsigned int*)od2->bits;
+    unsigned int *m2 = (unsigned int*)od2->mask;

-    for (i = 0; i < sizeof(*dest)/sizeof(unsigned long); i++)
+    for (i = 0; i < sizeof(*dest)/sizeof(unsigned int); i++)
      d[i] = (((s[i] & ~m1[i]) | (o1[i] & m1[i])) & ~m2[i]) | (o2[i] & m2[i]);
}
else
{
-    for (i = 0; i < sizeof(*dest)/sizeof(unsigned long); i++)
+    for (i = 0; i < sizeof(*dest)/sizeof(unsigned int); i++)
      d[i] = (s[i] & ~m1[i]) | (o1[i] & m1[i]);
}
}
@@ -562,7 +562,7 @@ EXPORT int MM_clientset(int action, Imod
   /* do these at least once */
#define cs (*((struct ClientSettings*)0))
#define ss (*((struct ShipSettings*)0))
-    assert((sizeof(cs) % sizeof(unsigned long)) == 0);
+    assert((sizeof(cs) % sizeof(uint32_t)) == 0);
   assert(COUNT(cs.long_set) == COUNT(long_names));
   assert(COUNT(cs.short_set) == COUNT(short_names));
   assert(COUNT(cs.byte_set) == COUNT(byte_names));

Net has some structs that it checks for size. I'm not sure why, but it's easy to fix:

Code: Show/Hide

--- src/include/net.h 2519ebab5906c86db40186b3053ded968e904ec4
+++ src/include/net.h 40bb87f50a61ba404a94ce879367f00a54334f2e
@@ -55,13 +55,13 @@ struct net_stats

struct net_stats
{
- unsigned long pcountpings, pktsent, pktrecvd;
- unsigned long bytesent, byterecvd;
- unsigned long buffercount, buffersused;
+ unsigned int pcountpings, pktsent, pktrecvd;
+ unsigned int bytesent, byterecvd;
+ unsigned int buffercount, buffersused;
#define NET_GROUPED_STATS_LEN 8
- unsigned long grouped_stats[NET_GROUPED_STATS_LEN];
+ unsigned int grouped_stats[NET_GROUPED_STATS_LEN];
#define NET_PRI_STATS_LEN 5 /* must match BW_PRIS in bwlimit.h */
- unsigned long pri_stats[NET_PRI_STATS_LEN];
+ unsigned int pri_stats[NET_PRI_STATS_LEN];
byte reserved[176];
};

@@ -70,9 +70,9 @@ struct net_client_stats
/* sequence numbers */
i32 s2cn, c2sn;
/* counts of stuff sent and recvd */
- unsigned long pktsent, pktrecvd, bytesent, byterecvd;
+ unsigned int pktsent, pktrecvd, bytesent, byterecvd;
/* count of s2c packets dropped */
- unsigned long pktdropped;
+ unsigned int pktdropped;
unsigned int unused1;
/* encryption type */
const char *encname;

rwlock.c needs to include asss.h for the #pragma pack(1) directive, otherwise some of the structs end up as a different size! This one took me a while to track down...

Code: Show/Hide

--- src/main/rwlock.c 172c3d5dcb2a2be709a77a1d990a044ba18f10b2
+++ src/main/rwlock.c 7199409e35ccd910fad5ac45da1b4c82bbdb82b8
@@ -29,6 +29,8 @@
* for write access can also lock it for read access any number of
* times; again it must be unlocked the same number.
*/
+#include "asss.h"
+
#include "pthread.h"
#include "errno.h"

Trash Talk :: RE: SubSpace Peer Protocol

K' — Tue, 06 Jan 2009 17:52:06 GMT

Author: K'
Posted: Tue Jan 06, 2009 12:52 pm (GMT -5)

Snrrrub, you're back in business now? I think last time I emailed you there was no response at the other end.

ASSS Questions :: RE: Loading pymod

Dr Brain — Tue, 06 Jan 2009 03:34:13 GMT

Author: Dr Brain
Posted: Mon Jan 05, 2009 10:34 pm (GMT -5)

The 64-bit problems run deep. I've made it a personal project to port asss to amd64, but you're better off looking into the 32-bit python stuff for now.

ASSS Questions :: RE: Broadcast

Hakaku — Tue, 06 Jan 2009 01:30:40 GMT

Author: Hakaku
Posted: Mon Jan 05, 2009 8:30 pm (GMT -5)

Bak wrote:

grep the source code for that message to figure out the conditions that trigger it

fg_wz.py states:

Code: Show/Hide

else:
   # couldn't place in original location, and couldn't place in
   # center either. this is really bad.
   lm.LogA(asss.L_ERROR, 'flagcore', a,
         ("failed to place a flag at (%d,%d)-%d, "
         "fallback disabled")% (cx, cy, r))

It's not very reassuring since there's nothing blocking the flags in the center of the map.

Edit: here's what I mean (dont mind my cont gfx though)

ASSS Questions :: RE: Broadcast

Bak — Mon, 05 Jan 2009 23:53:01 GMT

Author: Bak
Posted: Mon Jan 05, 2009 6:53 pm (GMT -5)

grep the source code for that message to figure out the conditions that trigger it

Trash Talk :: RE: SubSpace Peer Protocol

Bak — Mon, 05 Jan 2009 23:51:48 GMT

Author: Bak
Posted: Mon Jan 05, 2009 6:51 pm (GMT -5)

ohhh clever samp

ASSS Custom Projects :: RE: drone

Bak — Mon, 05 Jan 2009 23:45:57 GMT

Author: Bak
Posted: Mon Jan 05, 2009 6:45 pm (GMT -5)

addDrone and spawnDrone are interface functions, not asss commands. The only command used is ?drone, and the behavior is customizable by the specific drone type.

if drone is loaded and attached correctly (use "?lsmod -a" to check), load the simple module to spawn a drone (see simple.c in MM_LOAD for how to do this in code). use "?insmod simple" (or drone:simple or <dllname>:simple if you compiled it into a non-core .dll or .so)

that should spawn one drone which follows enemies and shoots at them (I think the bounty is the drone's energy).

the following behavior looks incomplete which is weird. If you want to be able to do "?drone follow" you'll need to code the droneAction function to populate the PerDroneData.playerToFollow field, although there's no notion of ownership with the simple bot

Trash Talk :: RE: SubSpace Peer Protocol

L.C. — Mon, 05 Jan 2009 22:04:51 GMT

Author: L.C.
Posted: Mon Jan 05, 2009 5:04 pm (GMT -5)

Quote:

I think it's to have multiple servers handle different arenas in a zone to split the load

Correct. This is probably how TW does or used to operate. ;o

ASSS Questions :: RE: Broadcast

Hakaku — Mon, 05 Jan 2009 20:20:18 GMT

Author: Hakaku
Posted: Mon Jan 05, 2009 3:20 pm (GMT -5)

Bah nevermind, it just wasn't listed in the userguide. Here's the full list if anyone needs it:

Quote:

seemodchat
sendmodchat
sendsoundmessages
uploadfile
seesysoplogall
seesysoplogarena
seeprivarena
seeprivfreq
bypasssecurity
broadcastbot
broadcastany
invisiblespectator
unlimitedchat
changesettings
isstaff
seeallstaff

On a non-bot related note, what could be causing the following error:

Code: Show/Hide

E <flagcore> {0} failed to place a flag at (512,512)-50, fallback disabled
E <flagcore> {0} failed to place a flag at (512,512)-50, fallback disabled

It seems to cause flags to spawn very awkwardly when the flags are reset (either when the arena is created or after a player wins the flaggame).

ASSS Questions :: Broadcast

Hakaku — Mon, 05 Jan 2009 19:11:42 GMT

Author: Hakaku
Subject: Broadcast
Posted: Mon Jan 05, 2009 2:11 pm (GMT -5)

When running a bot over ASSS, I keep getting flooded with the following error message :

Code: Show/Hide

M <objects> {0} [BotName] attempted rebroadcast without permission

According to objects.c, the bot would require the permission BROADCAST_BOT or BROADCAST_ANY in order to be able to toggle objects, but it doesn't state how to give a player/bot either of these permissions. Would anyone know how it's possible?

Trash Talk :: Akira (2011)

Sercit — Mon, 05 Jan 2009 16:05:44 GMT

Author: Sercit
Subject: Akira (2011)
Posted: Mon Jan 05, 2009 11:05 am (GMT -5)

The rumor mill is squawking that they're planning on doing another Akira for 2011. Dunno if it's going to be a redo, or what, but I can't imagine how they can top the original. I mean what makes the artistic nature of it so damn epic is that it's a late-80's anime and it still looks high-end for even the modern series that have come out now. It's a technical achievement beyond all else.

Any thoughts on it?

Trash Talk :: RE: SubSpace Peer Protocol

Samapico — Mon, 05 Jan 2009 15:45:00 GMT

Author: Samapico
Posted: Mon Jan 05, 2009 10:45 am (GMT -5)

I think it's to have multiple servers handle different arenas in a zone to split the load

Trash Talk :: RE: SubSpace Peer Protocol

k0zy — Mon, 05 Jan 2009 15:14:04 GMT

Author: k0zy
Posted: Mon Jan 05, 2009 10:14 am (GMT -5)

Okay... and why have the servers to communicate with each other? What's the benefit?

Trash Talk :: RE: SubSpace Peer Protocol

Doc Flabby — Mon, 05 Jan 2009 14:01:22 GMT

Author: Doc Flabby
Posted: Mon Jan 05, 2009 9:01 am (GMT -5)

Is this related to the ini file peer settings i found?

http://forums.minegoboom.com/viewtopic.php?t=7843

ASSS Custom Projects :: RE: drone

CaptainPoopface — Mon, 05 Jan 2009 12:11:29 GMT

Author: CaptainPoopface
Posted: Mon Jan 05, 2009 7:11 am (GMT -5)

Sorry for bump, but I would really like to get this working. It looks like you've put a great deal of work into it, and I want to play with it.

I cannot seem to spawn a drone. I've added the cmd_'s for addDrone, spawnDrone and removeDrone; put drone in the modules.conf AFTER objects, and referenced it in the arena.conf. The module attaches successfully. I try ?addDrone and ?spawnDrone (with no arguments) but nothing happens. What am I doing wrong?

Trash Talk :: RE: SubSpace Peer Protocol

Snrrrub — Mon, 05 Jan 2009 03:40:13 GMT

Author: Snrrrub
Posted: Sun Jan 04, 2009 10:40 pm (GMT -5)

This is by no means a secure protocol. But if your peers are running in the same subnet or you can make sure that nobody's intercepting messages along the way, it's sufficient.

Then again, if we're on the topic of security and man-in-the-middle attacks, no communication channel within SubSpace is secure so we might as well get the benefits of a stateless protocol and have some way of rejecting 99.99% of likely attacks.

So while it's not theoretically secure, it's certainly good enough for the purpose it needs to serve.

-Snrrrub

Trash Talk :: RE: SubSpace Peer Protocol

Bak — Mon, 05 Jan 2009 03:29:45 GMT

Author: Bak
Posted: Sun Jan 04, 2009 10:29 pm (GMT -5)

that sounds about right

ASSS Questions :: RE: Loading pymod

Dr Brain — Mon, 05 Jan 2009 03:08:35 GMT

Author: Dr Brain
Posted: Sun Jan 04, 2009 10:08 pm (GMT -5)

I've got things compiling, but there are some issues with stack corruption that I'm still investigating.

Trash Talk :: RE: SubSpace Peer Protocol

Dr Brain — Mon, 05 Jan 2009 02:48:13 GMT

Author: Dr Brain
Posted: Sun Jan 04, 2009 9:48 pm (GMT -5)

Isn't the CRC the same thing as the password? Knowing the CRC is enough to gain access to all of the functionality provided by the password. Seems a silly way to provide security.

Trash Talk :: RE: SubSpace Peer Protocol

Snrrrub — Mon, 05 Jan 2009 02:03:35 GMT

Author: Snrrrub
Posted: Sun Jan 04, 2009 9:03 pm (GMT -5)

Peers have to know about each other. Basically, you end up forming a trusted network based on a shared secret (the password).

To answer your question, yes, the CRCs are compared and the server admins are expected to share a common password for their peer network.

-Snrrrub

ASSS Questions :: RE: Loading pymod

Dr Brain — Mon, 05 Jan 2009 01:19:20 GMT

Author: Dr Brain
Posted: Sun Jan 04, 2009 8:19 pm (GMT -5)

I just started working on converting asss to 64-bit. There are some issues with type sizes. A lot of modules use "unsigned long" for no apparent reason. Maybe grel was worried about systems where an int is 2 bytes?

ASSS Custom Projects :: RE: Custom bomb trajectory

Dr Brain — Mon, 05 Jan 2009 01:16:04 GMT

Author: Dr Brain
Posted: Sun Jan 04, 2009 8:16 pm (GMT -5)

CaptainPoopface wrote:

Another workaround for disappearing the source player's regular bomb is to give him L1 bombs and make those invisible with 0 damage, or very short alive time. However, if you are "animating" a fake bomb using a series of velocitized bombs with short alive time, will they still have accurate collision as if they were a single bomb? It seems like you would have to dovetail the events carefully to avoid time gaps (possibly creating moments of double damage), and an object traveling faster than a ship and sending a lot of data sounds like a recipe for lag discrepancies. Is that right?

The primary problem isn't lag, but rather knowing when a bomb has detonated (and when to stop relaying it along its course).

Trash Talk :: RE: SubSpace Peer Protocol

Bak — Mon, 05 Jan 2009 01:03:55 GMT

Author: Bak
Posted: Sun Jan 04, 2009 8:03 pm (GMT -5)

is the crc32 password actually checked? how do they know the password of other zones?

LVZ/LVL Questions :: RE: dynamic runtime LVZ options

tcsoccerman — Mon, 05 Jan 2009 00:09:11 GMT

Author: tcsoccerman
Posted: Sun Jan 04, 2009 7:09 pm (GMT -5)

yes you can, i think theres a custom module that does that.