Next Update v4 - March 2016 Update is out!

[Mechanical Mind]

Any chance this could be looked at?
[url]https://facepunch.com/showthread.php?t=1329504&p=43021162&viewfull=1#post43021162[/url]

Replicating

] lua_run Entity(1):SetModelScale(8)
> Entity(1):SetModelScale(8)...
] lua_run print(Entity(1):OBBMaxs())
> print(Entity(1):OBBMaxs())...
128.000000 128.000000 576.000000
] lua_run Entity(1):SetHull(Vector(-16,-16,0) * 8, Vector(16,16,72)*8)
> Entity(1):SetHull(Vector(-16,-16,0) * 8, Vector(16,16,72)*8)...
] lua_run print(Entity(1):OBBMaxs())
> print(Entity(1):OBBMaxs())...
1024.000000 1024.000000 4608.000000

Using SetModelScale makes it impossible to get Hull and OBB consistent.

Note: Same issue as [url]https://github.com/Facepunch/garrysmod-issues/issues/2193[/url]

[Rubat]

[QUOTE=Stiffy360;49409171][URL="http://tf2maps.net/threads/repacking-your-maps.24822/"]TF2 added a cool thing[/URL] that compresses maps better than what bz2 compression does, and it works straight out of the box with garrysmod. However, this crashes the client on Macs. It works fine on PC. Any chance of getting that looked into? I can submit samples if need be.[/QUOTE]

It works because I made it work.

As I said:
[QUOTE=Robotboy655;49405296]when most crashes reports are of the sort "I got a crash" or "I got a crash, here's my console log, fix it now".[/QUOTE]

I need more info than "it doesnt work".

[Zeh Matt]

@Robotboy655

It would be really cool if we had a hook that NPCs use to decide their enemy, I guess GMod does something like most of the SDK mods where they just use the closest player for their FindEnemy function.

Also SetEnemy does not work on normal NPCs as their schedules use the FindEnemy function to keep looking for a new enemy and just overwrites what has been set by Lua, so only a hook would work I guess?

[Stiffy360]

[QUOTE=Robotboy655;49412328]It works because I made it work.

As I said:

I need more info than "it doesnt work".[/QUOTE]

I don't have a mac to test it on, or logs. I just got reports from players that it was. That's why I was offering samples for someone to test with.

[DrVrej]

[QUOTE=Zeh Matt;49412345]@Robotboy655

It would be really cool if we had a hook that NPCs use to decide their enemy, I guess GMod does something like most of the SDK mods where they just use the closest player for their FindEnemy function.

Also SetEnemy does not work on normal NPCs as their schedules use the FindEnemy function to keep looking for a new enemy and just overwrites what has been set by Lua, so only a hook would work I guess?[/QUOTE]

A lot of the NPC functions don't work anymore and most of the NPC functions aren't even ported over to Lua.

[Zeh Matt]

[QUOTE=vrej;49412543]A lot of the NPC functions don't work anymore and most of the NPC functions aren't even ported over to Lua.[/QUOTE]

Doesn't matter if you have NPC functions, this is just enemy decision, I already checked how its done, its just using the C function UTIL_GetLocalPlayer with a bool and vector passed to get the closest player, so this kind of lookup could be done with a hook.

Edit: To simplify why I need it, still working on my hl2 coop gamemode and the NPCs are just horrible in deciding who to face, so if I had the option to control this by Lua it would already help NPCs not fighting the same player because hes is the closest, simple as that really.

[dezamb]

Nevermind.

[Bo98]

[QUOTE=Stiffy360;49412368]I don't have a mac to test it on, or logs. I just got reports from players that it was. That's why I was offering samples for someone to test with.[/QUOTE]

I'll test it if you want.

[Zeh Matt]

[QUOTE=Robotboy655;49406241]I guess it's something to look into, it's supposed to be 4k as far as i can tell, there's another one at 2k, but its called dynamicmodels. I guess some preprocessor fuckery is going on.[/QUOTE]

qlimits.h:
#define MAX_MODEL_INDEX_BITS 11 // sent as a short
#define MAX_MODELS (1<<MAX_MODEL_INDEX_BITS)

Perhaps you forgot that?

[Rubat]

[QUOTE=Zeh Matt;49413827]qlimits.h:
#define MAX_MODEL_INDEX_BITS 11 // sent as a short
#define MAX_MODELS (1<<MAX_MODEL_INDEX_BITS)

Perhaps you forgot that?[/QUOTE]

No I haven't, as I already said on GitHub, I just tested it wrong and it will be raised to 4k.

[hootiegerbil]

This guy just came on my server and gave everyone the "I am the real garry achievement" but I think he hacked it. Is it real? [URL]http://steamcommunity.com/id/CommanderNZ[/URL]

[zamboni]

[QUOTE=hootiegerbil;49417380]This guy just came on my server and gave everyone the "I am the real garry achievement" is this the real one? [URL]http://steamcommunity.com/id/CommanderNZ[/URL][/QUOTE]

no.

[hootiegerbil]

The comments on his profile from people who think he's real... I'm going to be sick

[YourStalker]

That guy above you posted is [url=https://facepunch.com/member.php?u=240118]this[/url] guy.

[Zelpa]

[QUOTE=YourStalker;49418549]That guy above you posted is [url=https://facepunch.com/member.php?u=240118]this[/guy] guy.[/QUOTE]

excellent use of the unknown [guy] tag.

[The Commander]

[QUOTE=hootiegerbil;49417427]The comments on his profile from people who think he's real... I'm going to be sick[/QUOTE]
I was sick also because people couldn't look at the user name history or the Steam name to clearly see it's not Garry.

[Zeh Matt]

[QUOTE=Robotboy655;49414674]No I haven't, as I already said on GitHub, I just tested it wrong and it will be raised to 4k.[/QUOTE]

Excellent, also it would be cool if you could change the model loading into dynamic model loading, the stringtable is managed and once the ref count hits zero it would be deleted from the entry, that way 4k is sufficient at any given time unless you have for whatever insane reason 4k ents with unique models.

Sadly the Source SDK 2013 does not provide much information about it, however it seems as you simply have to register it as a dynamic model and wait for the callback that informs that the model data can be used.

You will probably figure it out.

[Dynn]

Has anyone else experienced errors like this on clients being on a server for more than 5-10 hours.
I got a large clientside error log filled with them and asking my players it seems to happen after they've been on the server for a very long(?) time.

Mickey #HOLIDAYS|249|STEAM_0:0:133286594


[ERROR] gamemodes/sandbox/entities/effects/selection_indicator.lua:57: attempt to compare nil with number
  1. unknown - gamemodes/sandbox/entities/effects/selection_indicator.lua:57


Mickey #HOLIDAYS|249|STEAM_0:0:133286594

[ERROR] gamemodes/sandbox/entities/effects/selection_ring.lua:57: attempt to compare nil with number
  1. unknown - gamemodes/sandbox/entities/effects/selection_ring.lua:57



Mickey #HOLIDAYS|249|STEAM_0:0:133286594

[ERROR] gamemodes/sandbox/entities/effects/selection_ring.lua:57: attempt to compare nil with number
  1. unknown - gamemodes/sandbox/entities/effects/selection_ring.lua:57



Mickey #HOLIDAYS|249|STEAM_0:0:133286594

[ERROR] gamemodes/sandbox/entities/effects/selection_ring.lua:57: attempt to compare nil with number
  1. unknown - gamemodes/sandbox/entities/effects/selection_ring.lua:57



Jyden|91|STEAM_0:1:56542945

[ERROR] gamemodes/base/entities/effects/tooltracer.lua:41: attempt to compare nil with number
  1. unknown - gamemodes/base/entities/effects/tooltracer.lua:41

[deinemudda32]

[QUOTE=ms333;49422183]Has anyone else experienced errors like this on clients being on a server for more than 5-10 hours.
I got a large clientside error log filled with them and asking my players it seems to happen after they've been on the server for a very long(?) time.

Mickey #HOLIDAYS|249|STEAM_0:0:133286594


[ERROR] gamemodes/sandbox/entities/effects/selection_indicator.lua:57: attempt to compare nil with number
  1. unknown - gamemodes/sandbox/entities/effects/selection_indicator.lua:57


Mickey #HOLIDAYS|249|STEAM_0:0:133286594

[ERROR] gamemodes/sandbox/entities/effects/selection_ring.lua:57: attempt to compare nil with number
  1. unknown - gamemodes/sandbox/entities/effects/selection_ring.lua:57



Mickey #HOLIDAYS|249|STEAM_0:0:133286594

[ERROR] gamemodes/sandbox/entities/effects/selection_ring.lua:57: attempt to compare nil with number
  1. unknown - gamemodes/sandbox/entities/effects/selection_ring.lua:57



Mickey #HOLIDAYS|249|STEAM_0:0:133286594

[ERROR] gamemodes/sandbox/entities/effects/selection_ring.lua:57: attempt to compare nil with number
  1. unknown - gamemodes/sandbox/entities/effects/selection_ring.lua:57



Jyden|91|STEAM_0:1:56542945

[ERROR] gamemodes/base/entities/effects/tooltracer.lua:41: attempt to compare nil with number
  1. unknown - gamemodes/base/entities/effects/tooltracer.lua:41

[/QUOTE]

Me too :D

[UselessGhost]

[url]https://twitter.com/FacepunchBot/status/685875231634882561[/url]

Whaaaaaat, willox is working on GMod now? :excited:

[zerf]

[QUOTE=Joeyl10;49493621][url]https://twitter.com/FacepunchBot/status/685875231634882561[/url]

Whaaaaaat, willox is working on GMod now? :excited:[/QUOTE]

Grats to my homeboy

[Teddi]

Grats Willox! Was amazed I got an email saying an issue I opened a year and a bit ago has been fixed :v:

[ARitz Cracker]

GMod may yet have a chance of surviving another 5 years.

[_Kilburn]

Hey guys, long time no see! Sorry for the really long abscence, my full time job has been really getting in the way for most of last year and I couldn't really spend much time on GMod at all. Things are starting to get a bit more relaxed though so I should be able to get things done again now.

I fixed a couple issues with NW2 vars, turns out they acted oddly when set on the world entity and that's precisely where global networked vars are stored. Can't really tell whether that's going to fix the random crashes or not though, I never managed to get them to crash on my end. Even by spawning hundreds of props on a dedicated server and setting a bunch of random strings on them every frame. :v:

The changes aren't on the dev branch yet because there are still a few minor details I'd like to fix but expect a small update tomorrow. Ideally I'd like to completely replace NW vars once we're 100% sure they're stable, and maybe DT vars as well, reverting the maximum DT var count back to 4 per type and internally using NW2 vars for higher indices. You wouldn't notice the difference and it would ensure backward compatibility with scripts that rely on the current 16 DT vars per type.

[UselessGhost]

[QUOTE=_Kilburn;49494280]
I fixed a couple issues with NW2 vars, turns out they acted oddly when set on the world entity and that's precisely where global networked vars are stored. Can't really tell whether that's going to fix the random crashes or not though, I never managed to get them to crash on my end. Even by spawning hundreds of props on a dedicated server and setting a bunch of random strings on them every frame. :v:[/QUOTE]

If I recall correctly the issue that was plaguing the new NWVars wasn't crashing, it was desyncs. I couldn't manage to get this to happen when I tested them a couple weeks ago, though.

[Grocel]

[QUOTE=_Kilburn;49494280]The changes aren't on the dev branch yet because there are still a few minor details I'd like to fix but expect a small update tomorrow. Ideally I'd like to completely replace NW vars once we're 100% sure they're stable, and maybe DT vars as well, reverting the maximum DT var count back to 4 per type and internally using NW2 vars for higher indices. You wouldn't notice the difference and it would ensure backward compatibility with scripts that rely on the current 16 DT vars per type.[/QUOTE]

Nice to see you back in action. Couldn't you just replace the internal DT stuff completely with NW2's system?

[Revenge282]

[QUOTE=Grocel;49495676]Nice to see you back in action. Couldn't you just replace the internal DT stuff completely with NW2's system?[/QUOTE]

Or just only use NW2?

[Giraffen93]

[QUOTE=_Kilburn;49494280]Hey guys, long time no see! Sorry for the really long abscence, my full time job has been really getting in the way for most of last year and I couldn't really spend much time on GMod at all. Things are starting to get a bit more relaxed though so I should be able to get things done again now.

I fixed a couple issues with NW2 vars, turns out they acted oddly when set on the world entity and that's precisely where global networked vars are stored. Can't really tell whether that's going to fix the random crashes or not though, I never managed to get them to crash on my end. Even by spawning hundreds of props on a dedicated server and setting a bunch of random strings on them every frame. :v:

The changes aren't on the dev branch yet because there are still a few minor details I'd like to fix but expect a small update tomorrow. Ideally I'd like to completely replace NW vars once we're 100% sure they're stable, and maybe DT vars as well, reverting the maximum DT var count back to 4 per type and internally using NW2 vars for higher indices. You wouldn't notice the difference and it would ensure backward compatibility with scripts that rely on the current 16 DT vars per type.[/QUOTE]

It crashed a lot for me when used while spawning, and new entities etc.. Otherwise nothing really

[_Kilburn]

[QUOTE=Grocel;49495676]Nice to see you back in action. Couldn't you just replace the internal DT stuff completely with NW2's system?[/QUOTE]

Well both have their pros and cons. DT vars are very minimalistic so the only thing you network is the value assigned to them. On the other hand NW2 vars need to network some extra information such as their type and the index of their name in the stringtable. Not to mention the stringtable lookup you have to perform on every read/write access.

I'd recommend using DT vars as usual for scripted entities and NW2 vars to network custom data on non scripted entities such as players or NPCs. Definitely going to reduce the max number of DT vars though, right now there are 32 of each type (excluding strings), which means 160 values that exist for every entity, most of which will most likely never be used at all.

[Willox]

Recent changes that are live on the dev branch:

[quote]
[b]bug fixes[/b]
- Replicated Lua ConVars now network to clients when they join the server (FCVAR_REPLICATED flag)
- Crash when entities with invalid physobjs were in the way of a rotating door is gone
- The console will no longer hide behind the main menu if you open it too early
- ai.GetScheduleID and ai.GetTaskID should now work, allowing for old-style NPCs to be made*
- file.Exists/file.IsDir are no longer super slow

[b]additions[/b]
- JavaScript errors in the console now show the file path and line number of the bad code
- PANEL:SetDrawLanguageID(bool) can be used to hide the language selection box in text entries when typing non-US/UK/whatever
[/quote]

*I haven't actually made any NPCs yet, but an example I got from the old wiki seems to work.