Hi, me again.
Would it be possible to execute the hook or a callback at the peak of the error stack, like xpcall does?

] lua_run function a() error'a' end function b() a() end function c() b() end function d() c() end xpcall(d, function() print(debug.traceback()) end)
> function a() error'a' end function b() a() end function c() b() end function d() c() end xpcall(d, function() print(debug.traceback()) end)...
stack traceback:
	lua_run:1: in function <lua_run:1>
	[C]: in function 'error'
	lua_run:1: in function 'a'
	lua_run:1: in function 'b'
	lua_run:1: in function 'c'
	lua_run:1: in function <lua_run:1>
	[C]: in function 'xpcall'
	lua_run:1: in main chunk

I'm writing a sentry integration and would like to walk the stack myself to grab upvalues and the like.

Hello, can you please compile the x64 build, cause I can't, I get a lot of errors and warnings while compiling it, building x86 is ok
https://pastebin.com/9b1dCpfw

If you attempt to require() the module in any situation with srcds, you'll get the following error stack -

[ERROR] unable to create a hook for HandleClientLuaError
  1. unknown - [C]:-1
   2. require - [C]:-1
    3. unknown - lua/autorun/server/luaerrortest.lua:1

Contents to reproduce is literally require("luaerror") on its' own. This seems to propagate one of two issues - it'll either cause a bootloop or crash upon a reload of the map.

As the x64bit beta branch is increasing in popularity, it would be very useful to be able to actually use it, while keeping track of errors with your wonderful module!

I personally have no experience with C++, so I really have no clue how difficult it would be, but I am hoping that it would become possible in the near future, as we plan to switch our servers onto the 64bit branch.

require "luaerror2"
unable to create a hook for HandleClientLuaError
1: (message = "unable to create a hook for HandleClientLuaError") [addons/gcompute/lua/gcompute/execution/local/gluaexecutioninstance.lua: 178]
2: ("d:\servers\garrysmod\garrysmod\lua\bin\gmsv_luaerror2_win32.dll", { table: 0x2d3df088 }, "unable to create a hook for HandleClientLuaError") [[C]: -1]
3: ("luaerror2", userdata: 0x2d3deef8) [[C]: -1]
4: (__param1 = "luaerror2") [[string "*"]: 0]
5: xpcall (function (message)) [[C]: -1]
6: Start (self = { GCompute.Execution.GLuaExecutionInstance: 0x2d3d91f0 }) [addons/gcompute/lua/gcompute/execution/local/gluaexecutioninstance.lua: 173]
7: CreateExecutionInstance (self = { GCompute.Execution.GLuaExecutionContext: 0x2d3c8668 }, code = "require "luaerror2"", sourceId = "@repl_0", instanceOptions = 6, callback = nil) [addons/gcompute/lua/gcompute/execution/executioncontext.lua: 56]
8: HandleExecutionInstanceCreationRequest0 (self = { GCompute.Execution.RemoteExecutionContextHost: 0x2d3ca0f8 }, connection = { GLib.Net.Connection: 0x2d3d4548 }, inBuffer = { GLib.Net.Layer1.PinnedNetInBuffer: 0x2d3d6f68 }) [addons/gcompute/lua/gcompute/execution/remote/remoteexecutioncontexthost.lua: 94]
9: (self = { GCompute.Execution.RemoteExecutionContextHost: 0x2d3ca0f8 }, connection = { GLib.Net.Connection: 0x2d3d4548 }, inBuffer = { GLib.Net.Layer1.PinnedNetInBuffer: 0x2d3d6f68 }) [addons/gcompute/lua/gcompute/execution/remote/remoteexecutioncontexthost.lua: 78]
10: xpcall (GLib.Error) [[C]: -1]
11: () [addons/glib/lua/glib/threading/thread.lua: 262]

Getting this error when i require luaerror on 32 bit dedicated linux server unable to obtain AdvancedLuaErrorReporter happens on 1.5.8,1.5.7, 1.5.6

Getting a bootloop with this module loaded at startup.

Error as follows -

[ERROR] unable to sigscan function HandleClientLuaError
  1. unknown - [C]:-1
   2. require - [C]:-1
    3. unknown - snip
     4. include - [C]:-1
      5. IncludeServerFiles - snip
       6. unknown - snip
        7. include - [C]:-1
         8. unknown - snip
          9. DeriveGamemode - [C]:-1
           10. unknown -snip

The hook argument count is getting massive and the code a little messy.

There's a great opportunity here to squash together the LuaError and ClientLuaError hooks into a single one which reports the source of the error with an enum for example. Maybe move all these arguments into a table as well.

These changes should bump the major version up since they're backwards incompatible.

Would you be able to add some release binaries the the reop ( Win and Linux )? I'd love to use this lib but I have no idea how to build it.

Been finding out my server suddenly crashed the moment a map change happened, which seems to be related the moment another addon calls require("gm_luaerror"). For some reason, it does not affect it when the server process starts for the first time, only on map changes.

The server is running on the master branch.
The host machine runs on Windows Server 2019.

ClientLuaError returns completely random, seen it return basically anything including prop_physics entities.

Is support for x86-64 architecture planned?

Attempted to load this and wasn't working. I performed this test

require "luaerror"
hook.Add("LuaError","test", function(fullerror)
print("ooooof")
end)

It returned with

[ERROR] <color=255,100,100>LuaFoxx:1: Module not found!
stack traceback:
[C]: in function 'require'
<color=255,100,100>LuaFoxx:1: in main chunk
[C]: in function 'LUADEV_EXECUTE_FUNCTION'

I used the "gmsv_luaerror_linux.dll" and when using require("luaerror") when module placed in lua/bin/ it errors. When I down grade to the version from 2 years ago it now works.
Version: 1.5.3

I downloaded the cl version of the binary hoping to hide errors from the server somehow.
I required it in the client state not the menu state.

luaerror.EnableCompiletimeDetour(true) makes include think the file is empty or something.
I have tried having both of them false one true and both true.

Running script a.lua...
Couldn't include file 'a.lua' (File not found) ()

hook.Add("LuaError","heck",function(...)
print("oh no an error ",...)
end)
hook.Add("ClientLuaError","heck",hook.GetTable()["LuaError"]["heck"])

to trigger an error i am doing:
lua_run_cl -
lua_run_cl wat="
lua_run_cl local n n()

none of them print anything

Tomorrow's GMod update includes changes to the C++ API and/or patches some exploits. At any rate, the update has broken the gmsv_luaerror module:

[ERROR] unable to sigscan function HandleClientLuaError
  1. unknown - [C]:-1
   2. require - [C]:-1
    3. unknown - addons/path/to/my/lua/file.lua:XX

I don't really know much beyond that. Simply require'ing the module will cause the error. I know myself and several others still rely on this module, so if you'd be willing to test out the prerelease branch today or the main branch when the update comes out tomorrow, we'd really appreciate it.

With the most recent release (2021.01.22), an error followed by a crash is recorded when using the latest luaerror.

[ERROR] unable to obtain AdvancedLuaErrorReporter
  1. unknown - [C]:-1
   2. require - [C]:-1

I just can't use gmsv_luaerror_linux.

lua_run print(pcall(require, "luaerror"))
> print(pcall(require, "luaerror"))...
false   Couldn't load module library!

Using gmcl_luaerror_linux works.
Can someone tell me why?

Thanks for adding the workshop support!
Would you be able to expose the FindOwnerWorkshopAddon function as a Lua function please?
I'm doing stacktrace analysis and it'd be very handy to be able to tag each entry as the workshop addon rather than "unknown".
Thanks again

Both hooks add the stack object, however they seem to be slightly different. I can't seem to get the short_src from the ClientLuaError hook stack at all.

Clientside example:

Serverside example:

Yet server side seems to have the short_src.

Might be intentional or just a limitation by gmod, in that case feel free to close it.

Pending the next update, the module will load as intended but will fail to catch any errors. This can be confirmed on the latest dev build.

I downloaded the latest windows artifact from azure pipelines: https://metamann.visualstudio.com/GitHub%20danielga/_build/results?buildId=112&view=results

Running gmod without any beta branches on a Windows 10 desktop.

I simply get this error when trying to require luaerror, not much else to go on.

] lua_run_cl error("cl err")

[ERROR] LuaCmd:1: cl err
  1. error - [C]:-1
   2. unknown - LuaCmd:1

[Lexi|10|STEAM_0:1:16678762] Lua Error:

[ERROR] LuaCmd:1: cl err
  1. error - [C]:-1
   2. unknown - LuaCmd:1


] lua_run error("sv err")
> error("sv err")...

[ERROR] lua_run:1: sv err
  1. error - [C]:-1
   2. unknown - lua_run:1

] lua_run luaerror.EnableCompiletimeDetour(true)
> luaerror.EnableCompiletimeDetour(true)...
] lua_run_cl error("cl err")

[ERROR] LuaCmd:1: cl err
  1. error - [C]:-1
   2. unknown - LuaCmd:1

[Lexi|10|STEAM_0:1:16678762] Lua Error:

[ERROR] LuaCmd:1: cl err
  1. error - [C]:-1
   2. unknown - LuaCmd:1


] lua_run error("sv err")
> error("sv err")...

[ERROR] lua_run:1: sv err
  1. error - [C]:-1
   2. unknown - lua_run:1

] lua_run luaerror.EnableRuntimeDetour(true)
> luaerror.EnableRuntimeDetour(true)...
] lua_run_cl error("cl err")

[ERROR] LuaCmd:1: cl err
  1. error - [C]:-1
   2. unknown - LuaCmd:1

[Lexi|10|STEAM_0:1:16678762] Lua Error:

[ERROR] LuaCmd:1: cl err
  1. error - [C]:-1
   2. unknown - LuaCmd:1


] lua_run error("sv err")
> error("sv err")...

[ERROR] LuaCmd:1: cl err
  1. error - [C]:-1
   2. unknown - lua_run:1

The stack trace is correct:

] lua_run function test2() error("what?") end function mytest() test2() end mytest()
> function test2() error("what?") end function mytest() test2() end mytest()...

[ERROR] LuaCmd:1: cl err
  1. error - [C]:-1
   2. test2 - lua_run:1
    3. mytest - lua_run:1
     4. unknown - lua_run:1

And the hook gets the right value:

] lua_run hook.Add("LuaError", "wut", function(_, e) print(e) end)
> hook.Add("LuaError", "wut", function(_, e) print(e) end)...
] lua_run error("sv err")
> error("sv err")...
lua_run:1: sv err

[ERROR] LuaCmd:1: cl err
  1. error - [C]:-1
   2. unknown - lua_run:1

it just sends lua client errors even as both are detoured, the same code works fine on a srcds server running on windows.

tested both on debian and ubuntu server

recompile it to fix it dad

Just started getting this recently. The first time a client receives an error with this module loaded, the server immediately segfaults.

Segmentation fault (core dumped)
Add "-debug" to the ./srcds_run command line to generate a debug.log to help with solving this problem

I know nothing about debugging but here's the initial output from gdb as per the guide here. I'm not sure how to get more valuable info from it.

Reading symbols from srcds_linux...
(No debugging symbols found in srcds_linux)
BFD: warning: /home/avtomat/serverfiles/core has a segment extending past end of file
[New LWP 3026220]
[New LWP 3027235]
[New LWP 3027234]
[New LWP 3027224]
[New LWP 3026223]
[New LWP 3026222]
[New LWP 3027222]
[New LWP 3026224]
[New LWP 3027226]
[New LWP 3027220]
[New LWP 3027236]
[New LWP 3026246]
--Type <RET> for more, q to quit, c to continue without paging--
[New LWP 3027229]
[New LWP 3027227]
[New LWP 3026247]
[New LWP 3026249]
[New LWP 3027230]
[New LWP 3026248]
[New LWP 3027219]
[New LWP 3026744]

warning: Error reading shared library list entry at 0x64646436
Cannot access memory at address 0x61353069
Cannot access memory at address 0x61353065
Failed to read a valid object file image from memory.
Core was generated by `./srcds_linux -game garrysmod -ip 0.0.0.0 -port 27016 +host_workshop_collection'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0  0xf041b72e in ?? ()
[Current thread is 1 (LWP 3026220)]

No debug.log was generated.

When a workshop addon throws an error, the name of the addon is reported in the console like this:

[Kaboom - The Broken Addon] lua/autorun/server/boom.lua:1: this is an addon error

But this module reports it on its own:

lua/autorun/server/boom.lua:1: this is an addon error

The stack traces also lack any way of knowing what addon was responsible.
Is this fixable?
Thanks

Exe build: 20:02:11 Dec 14 2021 (8438) (4000)
GMod version 2021.12.14, branch: unknown
Windows 32bit Dedicated Server

If an error is thrown before ISteamHTTP is available (e.g. close to or on server boot), the server will hang and endlessly consume RAM until the server echos not enough memory.

Example error trace:

[ERROR] HTTP failed - ISteamHTTP isn't available!
  1. Post - lua/includes/modules/http.lua:71
   2. v - gamemode/path/that/hooks
    3. unknown - lua/includes/modules/hook.lua:96
     4. HTTP - [C]:-1

While currently correct >99% of the time, it's technically possible to have a source file with a : in it.

The current implementation appears to split the string by all :'s, and assumes the first element is the File, and the second is the Line Number.

Instead, to cover all cases, it could split by only the last : and make the same assumptions.

As the title says, when I require("luaerror") I'm receiving the error in the title, alongside my server crashing immediately. Server is running on Windows Server 2019, GMod version is the latest version of the main branch. I am on the latest version of luaerror.

This only started happening after I updated the GMod server today (first time in around 3 months), if it helps.

My other installed modules:
https://github.com/Metastruct/gm_enginespew
https://github.com/FredyH/GWSockets
https://github.com/WilliamVenner/gmsv_reqwest

When any error occurs on the server side, the server crashes

MDMP File Download
To watch web - https://dumps.metastruct.uk.to/analyze/34f26264bce2375373acb7a269422c491f08908d5e4d091f03a64a91f3123d13

The same as danielga/gm_filesystem#6 but for gm_luaerror

Hi there.
Using the most recent release, loading the module on my Ubuntu 16.04 server throws the following errors: