Something about msecval

[strelok]

Why, when I use this code (by Count Floyd):

if (pBot->msecdel <= gpGlobals->time)
{
pBot->msecdel = gpGlobals->time + 0.5;
if (pBot->msecnum > 0)
pBot->msecval = 450.0/pBot->msecnum;
pBot->msecnum = 0;
}
else
pBot->msecnum++;
if (pBot->msecval < 1) // don't allow msec to be less than 1...
pBot->msecval = 1;

if (pBot->msecval > 100) // ...or greater than 100
pBot->msecval = 100;

my bots freeze (in the beginning of a round) for some seconds?

[Pierre-Marie Baty]

This code is Rich "TheFatal" Whitehouse's method to estimate the duration of the next frame, in order to predict how long the pfnRunPlayerMove call for the movement of the bots should last. It looks correct, I find it strange that it could cause you trouble. Are you sure your problem comes from here ?

Anyway, there are several other methods to estimate the frame duration. The simplest one is to use the time difference between previous_time and gpGlobals->time in StartFrame(). But it is best to filter this value with time, and that's the purpose of msecnum and msecdel.

[koraX]

PMB is right, This piece of code should not cause errors. I'm using it too.

[botmeister]

Well, I've had poor results with this code, but I don't think it is the code that's the problem.

For example, I have two dedicated servers, both ar erunning realbot WIP#6 (and previously RB:AI ver 1.0). One server has the bots running around just slightly faster than they should, the other server has the bots running WAY faster than they should (2 or 3 x faster).

When I tried adding my own prototype bot into a game, I got the same problem - a bot that moved around way too fast depending on the server.

What i ended up doing, is setting the max speed of the bot entity to sv_maxspeed and that fixed the problem. I've added this "fix" to my bot manager (to be released soon), but a real solution to the problem should be found.

I posted this problem in botmans forum but no one had anything to offer at the time.

[Pierre-Marie Baty]

botmeister, your method will force the bots to move as fast as can be, whether they carry heavy weapons or not. Have you tried hooking pfnSetClientMaxSpeed() and set each bot's max speed to this value ?

[Bert]

I hook pfnSetClientMaxSpeed() and my bots move at the correct speed. I also use The Fatal's msec code, and I haven't noticed any problems with it.

[Cheeseh]

Isn't this not the problem in CS where the bots angles overflow when they spawn? You should know all about that PM :D

edit: may aswell explain some:

If you are using botmans template, if the bots try to move more than 10 degrees (I think, maybe its 20) they stop and try to turn. If they're angles have overflown this might jumble up the bots interpretation and are not turning properly, meaning they will never turn the 10 degrees (or whatever) and will not move.

Use BotFixIdealAngles() function when a bot spawns.

.. or is It IdealYaw()...

Something like that... I don't use those functions anymore :D

[Pierre-Marie Baty]

<offtopic>
how come there are a lot of bot developers in here and we never heard of their bots ? o_O
@Bert: You have an URL for your bot ? :)
</offtopic>

[Pierre-Marie Baty]

@Cheeseh: Strelok is using Count Floyd's code, and CF obviously fixed this angle problems for quite a time already :)

[Cheeseh]

Alright, well maybe they are just stopping to buy stuff eh? :p I haven't looked at CF's code, if you can debug, break when their move speed is 0 and see whats making it 0.

[Bert]

[Off-topic]
I think there are alot of tiny bots out there on the net, afterall with Botman's template code and readme, it's amazingly easy to get a basic bot in a mod.
As for my bot, I released the first version a month ago or so, but it's nothing more than a modified HPBbot. It's for the mod Science and Industry (www.planethalflife.com/si (http://www.planethalflife.com/si)) and it's more like a "Can I pull it off" release. There's some errors in the logic (Bots really like to kill scis) and it only comes with waypoints for 1 map. And it's not fully compatible with the latest version (97b). Try it if you like, but don't expect too much http://users.pandora.be/piele/MBFbot.zip . These days I seem to be stuck in an endless redesign cycle, I simply can not decide on what's the best way to handle things, so I keep changing stuff around :/
[/Offtopic]

[botmeister]

botmeister, your method will force the bots to move as fast as can be, whether they carry heavy weapons or not. Have you tried hooking pfnSetClientMaxSpeed() and set each bot's max speed to this value ?
Thanks again, I'll give this one a try.

[Pierre-Marie Baty]

@Bert: Okay, well, good luck, and btw when you come up with a "production release" of your bot we can make you part of Bots United perhaps (website, forum, etc.) :)

@botmeister: I find it curious that you never did. It's the way to do things, all CS bots hook pfnSetClientMaxSpeed() actually ! How come you never tried that ?

[botmeister]

I find it curious that you never did. It's the way to do things, all CS bots hook pfnSetClientMaxSpeed() actually ! How come you never tried that ?

I never did because I have not built much of a bot. Aside from the server management stuff, what I've done is enhanced existing bots which only goes so far.

I'm not sure how you would use the function just yet
void (*pfnSetClientMaxspeed) (const edict_t *pEdict, float fNewMaxspeed);

Can you give me a hint or two to move things along?

[Pierre-Marie Baty]

in your array of bots, have a variable for each bot to store its current max speed. When pfnSetClientMaxspeed() is called upon a bot entity, send fNewMaxspeed to this variable. And when pfnRunPlayerMove() happens for the same bot, use this variable as the move_speed and strafe_speed parameter (if you want to make the bot run or strafe full speed, else divide it by two.)

[botmeister]

in your array of bots, have a variable for each bot to store its current max speed. When pfnSetClientMaxspeed() is called upon a bot entity, send fNewMaxspeed to this variable. And when pfnRunPlayerMove() happens for the same bot, use this variable as the move_speed and strafe_speed parameter (if you want to make the bot run or strafe full speed, else divide it by two.)
Ok, from what I understand, pfnSetClientMaxspeed() is used to set the speed of a bot. What I'd need it for is to get the speed of a bot, since I do not know what the bots speed should be. So, are you saying that I hook onto the function, and whenever it gets called I grab whatever was passed to the function as the bots new speed?

Anyhow, what I'm doing is simply passing on everything I get from the pfnRunPlayerMove() hook which was called by the actual bot mod, so I don't understand why on some servers the bots run around real fast. I'll have another look at my code, perhaps there's a problem with passing on the same speeds all the time.

[strelok]

[QUOTE=Are you sure your problem comes from here ?[/QUOTE]
And if I use this code:

if (pBot->msecdel <= gpGlobals->time) {
pBot->msecdel = gpGlobals->time + 1.0;
if (pBot->msecnum > 0)
pBot->msecval = 800/pBot->msecnum;
pBot->msecnum = 0;
}
else
pBot->msecnum++;
if (pBot->msecval < 1)
pBot->msecval = 1;
if (pBot->msecval > 100)
pBot->msecval = 100;

bots move slowly (I with primary weapon run faster, than they with a knife). Also do not freeze.

[Pierre-Marie Baty]

@ botmeister: each time a player switches to a new weapon, his run speed changes. The function that the MOD code calls in the engine to change the client's max speed is SetClientMaxSpeed(). So yes, grab the value that is passed, and use it for your bots ; i.e. do this:

void pfnSetClientMaxspeed (const edict_t *pEdict, float fNewMaxspeed)
{
int client_index = ENTINDEX ((edict_t *) pEdict) - 1; // get client index

// is this message for a bot ? if so, remember its max speed
if (players[client_index].is_racc_bot)
players[client_index].Bot.BotMove.f_max_speed = fNewMaxspeed;

(*g_engfuncs.pfnSetClientMaxspeed) (pEdict, fNewMaxspeed);
}


@ Strelok: it is *normal* that the msec code has some influence on the bot's speed, but your problem doesn't come from here, because the msec code is *correct*. You should not change it. Instead, check for the bot's max speed. There is a problem in CS 1.6 where the player's max speed isn't set at round start (max_speed is initially 0 because pfnSetClientMaxSpeed() isn't called at round start like in 1.5) but it becomes set only when players switch to a new weapon. It may explain why your bots don't move as long as they don't change weapons.

[strelok]

2 Pierre-Marie Baty: It not a problem cs1.6. In cs1.5 all too most, but time of freezing is longer.

[Pierre-Marie Baty]

curious...
does it occur also if you set the mp_freezetime CVAR to 0 ?

[strelok]

NO:) If mp_freezetime = 0.0 all is normal.

[Pierre-Marie Baty]

Check what is the bot's max speed DURING the freeze time and AFTER it... that might be the problem, somehow.

[strelok]

max speed DURING the freeze time: cs1.5 = 1, cs1.6 = 320, AFTER it: cs1.5 = 250, cs1.6 = 320

[Pierre-Marie Baty]

Okay guys, we have an issue here. 320 is NOT a valid client speed in Counter-Strike. In Half-Life, it is.

Counter-Strike 1.6 does NOT send their max speeds to clients anymore. It's useless for us to hook pfnSetClientMaxSpeed() anymore. We have to fix the bot's max speed OURSELVES.

I suggest having a look in eLiTe's TeamBot code. It's REALLY old stuff, but there is interesting stuff. Because that's how the early bots were setting their max speeds before the pfnSetClientMaxSpeed() was discovered.

// This function is redundant, but the switch statement could be useful in future...
void SetSpeed (bot_t *pBot)
{
// This changes the bot's maximum speed depending on the weapon it is
// currently using

switch (pBot->current_weapon.iId)
{
case CS_WEAPON_P228:
pBot->f_max_speed = 250;
break;
case CS_WEAPON_SCOUT:
pBot->f_max_speed = 260;
break;
case CS_WEAPON_HEGRENADE:
pBot->f_max_speed = 260;
break;
case CS_WEAPON_XM1014:
pBot->f_max_speed = 240;
break;
case CS_WEAPON_C4:
pBot->f_max_speed = 250;
break;
case CS_WEAPON_MAC10:
pBot->f_max_speed = 250;
break;
case CS_WEAPON_AUG:
pBot->f_max_speed = 240;
break;
case CS_WEAPON_SMOKEGRENADE:
pBot->f_max_speed = 260;
break;
case CS_WEAPON_ELITE:
pBot->f_max_speed = 250;
break;
case CS_WEAPON_USP:
pBot->f_max_speed = 250;
break;
case CS_WEAPON_GLOCK18:
pBot->f_max_speed = 250;
break;
case CS_WEAPON_AWP:
pBot->f_max_speed = 210;
break;
case CS_WEAPON_MP5NAVY:
pBot->f_max_speed = 250;
break;
case CS_WEAPON_M249:
pBot->f_max_speed = 220;
break;
case CS_WEAPON_M3:
pBot->f_max_speed = 230;
break;
case CS_WEAPON_M4A1:
pBot->f_max_speed = 230;
break;
case CS_WEAPON_TMP:
pBot->f_max_speed = 250;
break;
case CS_WEAPON_G3SG1:
pBot->f_max_speed = 200;
break;
case CS_WEAPON_FLASHBANG:
pBot->f_max_speed = 260;
break;
case CS_WEAPON_DEAGLE:
pBot->f_max_speed = 250;
break;
case CS_WEAPON_SG552:
pBot->f_max_speed = 235;
break;
case CS_WEAPON_AK47:
pBot->f_max_speed = 221;
break;
case CS_WEAPON_KNIFE:
pBot->f_max_speed = 260;
break;
case CS_WEAPON_P90:
pBot->f_max_speed = 250;
break;
case CS_WEAPON_UMP45:
pBot->f_max_speed = 250;
break;
case CS_WEAPON_FIVESEVEN:
pBot->f_max_speed = 250;
break;
case CS_WEAPON_SG550:
pBot->f_max_speed = 210;
break;
}
}

We have to apply this each time a bot changes weapons. Eventually it would be good to check if these old values are still correct after 2 or 3 Counter-Strike updates. It's now the only way to fix the bot's "speed hack" bugs.

[botmeister]

We have to apply this each time a bot changes weapons. Eventually it would be good to check if these old values are still correct after 2 or 3 Counter-Strike updates. It's now the only way to fix the bot's "speed hack" bugs.
I knew something was going wrong. I had problems with the msec system, and the hook onto pfnSetClientMaxSpeed() seemed to correct the speed hack problem but I've noticed that the bots were still running around too fast. Also, I tried using sv_maxspeed but the cvar is not always initialized correctly. There's lots of bugs in CS 1.6 surrounding the speeds and my bet is that they broke the thing trying to hammer in the shield.

[KaCaT]

I use pfnSetClientMaxSpeed() for my bot in CS 1.5 and pev->maxspeed in CS 1.6. They are working well for me :)

[stefanhendriks]

I just get freaking nuts of CS 1.6 guys! Or should i say, get sick of this STEAM? Argh.

[Pierre-Marie Baty]

I use pfnSetClientMaxSpeed() for my bot in CS 1.5 and pev->maxspeed in CS 1.6. They are working well for me :)pEdict->v.maxspeed ? Hey, great ! How come we didn't think of it earlier ? wow, gotta credit KaCaT in my code now :D

I knew something was going wrong. [...] There's lots of bugs in CS 1.6 surrounding the speeds and my bet is that they broke the thing trying to hammer in the shield.No, actually I believe they did it on purpose to lower the network traffic. If you think about it it's completely stupid to send their max speeds to the clients over the network since the clients already know it (they're running the same game DLL, don't they ?) And since some players have the bad habit to switch weapons like crazy, I think it's wise they did so.
There are still a lot of things that could be removed from the network code. The CS netcode isn't one of the most clever ones around. Maybe TurtleRocker could tell us more about it.

[botmeister]

OK, I agree PMB, it is good idea to optimize the netcode and they probably did remove it on purpose. However I still don't understand why sv_maxspeed was being initialized incorrectly.

[stefanhendriks]

i don't get it either, but if pEdict->v.maxspeed works i have said nothing ;)

[Pierre-Marie Baty]

"sv_maxspeed was not initialized correctly" because sv_maxspeed is a server CVAR, not a client one, which means that if the game DLL chooses to rely on this value for the players' movements, ALL the players will run at the same speed all the time. This can't be, because CS lets players have different run speeds given the weapon they are carrying. That's why the game DLL of Counter-Strike does NOT rely on sv_maxspeed, but let the clients adapt their max speeds individually, unlike other MODs which don't feature different player run speeds.

[strelok]

2 Pierre-Marie Baty (http://forums.bots-united.com/member.php?u=2) vbmenu_register("postmenu_4927", true);
You function (void SetSpeed (bot_t *pBot)) only for CS 1.5? Where weapon FAMAS, GALIL?

[Pierre-Marie Baty]

right, it's not "my" function, really ; it's eLiTe's function (which he had in his TEAMbot). You'll have to figure out yourself what are the player's max speed when you are carrying a FAMAS or a GALIL.

But I suggest you rather use pBot->pEdict->v.maxspeed like KaCaT said.

[Pierre-Marie Baty]

BTW, KaCaT, does pEdict->v.maxspeed also work for CS 1.5 ?

[KaCaT]

No, it doesn't works for CS 1.5. The value it gives for bots in CS 1.5 always is 1000 8o

[Pierre-Marie Baty]

then how about this instead:

void pfnSetClientMaxspeed (const edict_t *pEdict, float fNewMaxspeed)
{
// update this client's max speed (Counter-Strike 1.5 doesn't do it right)
((edict_t *) pEdict)->v.maxspeed = fNewMaxspeed;
(*g_engfuncs.pfnSetClientMaxspeed) (pEdict, fNewMaxspeed);
}

:)

[strelok]

With a speed everything is all right. Thanks. :)
But the problem with freezing has remained. :(

[strelok]

What value should be up to freeze time and what after?

[Pierre-Marie Baty]

the bot's maxspeed should be kept at 0 before freeze time, and raised to the normal value (pBot->pEdict->v.maxspeed in CS 1.6 or f_max_speed passed in pfnSetClientMaxSpeed() in CS 1.5) immediately after it. But I'm not sure your problem comes from here...