| 00:20.18 | Noshei | yeah I read that |
| 00:20.49 | Noshei | that is a very interesting security lapse and I'm surprised blizzard hasn't fixed it yet |
| 00:21.38 | Ivnedar | they did add in that popup if you try to run code from the console now |
| 00:21.41 | Ivnedar | one off |
| 00:22.12 | Noshei | I don't remember ever getting that, and I've ran code from the console |
| 00:22.29 | Noshei | but that isn't really fixing the issue |
| 00:22.29 | Ivnedar | when did you try it last? |
| 00:22.37 | Noshei | yesterday |
| 00:24.12 | Noshei | From what I've read, people think the RemoveExtraSpaces function could be made local, instead of global. pretty easy fix |
| 00:31.30 | Imhothar | Then you break any addon that uses RemoveExtraSpaces for its own purpose |
| 00:32.22 | Imhothar | There isn't really any easy fix as long as the chat UI is fed from Lua |
| 00:35.18 | Ivnedar | you'd need to treat stuff coming in /run differently to from an addon |
| 00:35.33 | Imhothar | in what way |
| 00:35.42 | Imhothar | that doesn't make /run useless |
| 00:36.18 | Ivnedar | could decide that some actions are malicious enough to block, lick assigning anything to RunScript |
| 00:36.28 | Ivnedar | like even |
| 00:37.40 | Ivnedar | I'm sure it'd turn into an arms race of people finding creative ways around it |
| 00:38.38 | Imhothar | A simple lexical test wouldn't do it. You'd have to actively track the *real* value of RunScript throughout all addon running code so it's not shadowed by something or has an alias or whatever |
| 00:39.17 | Imhothar | _G["R".."unScript"] |
| 00:39.21 | Ivnedar | :) |
| 00:40.53 | Imhothar | so in order for that to work you need 1. Know the value of the original RunScript function and 2. Reject any /run script that loads that value |
| 00:41.06 | Imhothar | which implies you need to execute /run opcode for opcode |
| 00:41.59 | Ivnedar | yeah, if the action came from console and it's referencing a protected function's address, prompt |
| 00:42.58 | Imhothar | Well, then you just replace some arbitrary function that is guaranteed executed at some point during gameplay |
| 00:43.20 | Imhothar | and give it a closure that contains RunScript but isn't called until said hook runs |
| 00:43.33 | Imhothar | in a context that is no longer /run |
| 00:44.49 | Imhothar | it's a very one-sided arms race |
| 00:45.18 | Imhothar | Lua is just too damn flexible to reliably prevent this |
| 00:47.34 | Ivnedar | I assume there'd be a big unroar if they gave console and addon different global namespaces |
| 00:48.30 | Imhothar | I put that into the "make /run useless" box |
| 00:48.47 | Imhothar | since it's primary use is addon development/debugging |
| 00:52.15 | Imhothar | But it's an interesting approach if supported from the get-go. Factorio creates a new Lua instance for every mods and they can only communicate via a dedicated RPC interface |
| 00:52.16 | Ivnedar | config.wtf option to enable /run |
| 00:52.39 | Ivnedar | people who want to tinker can enable it, general population are somewhat protected |
| 00:52.43 | Imhothar | which means /script has no access to a mod's data |
| 00:53.33 | Imhothar | the separation is cool from a "call your things whatever you want" perspective but not having access to the stuff in /script can be a pain |
| 00:53.37 | Ivnedar | have another slash command that can read mod data |
| 00:54.07 | Ivnedar | or even have /run switch behaviour based on the toggle in config.wtf |
| 00:55.33 | Ivnedar | can lua make access to globals readonly in some way? |
| 00:56.19 | Imhothar | no |
| 00:56.27 | Imhothar | well |
| 00:56.49 | Imhothar | you can give _G a metatable where the mutating operations don't do anything |
| 00:57.02 | Imhothar | but that doesn't apply transitively |
| 00:57.24 | Imhothar | and anyone can change the metatable at any point so... |
| 00:57.47 | Ivnedar | unless that ability is removed |
| 00:58.08 | Imhothar | it still only protects the first level of the global table |
| 00:58.26 | Imhothar | so anything that is nested needs the same metatable |
| 00:58.34 | Imhothar | and the same protection |
| 00:58.44 | Imhothar | which easily becomes a nightmare to implement |
| 00:58.55 | Ivnedar | you'd need some form of metatable proxy that had no way of changing the metatable it's referring to |
| 00:59.19 | Ivnedar | overlay that can read values from below, but not alter them |
| 00:59.25 | Imhothar | I often wonder why WoW/Rift don't use the userdata type |
| 00:59.58 | Imhothar | userdata values are opaque values representing something in C that Lua doesn't have access to |
| 01:00.10 | Imhothar | and you can change its metatable only from the C interface |
| 01:00.13 | Ivnedar | oops, meeting, bbl |
| 01:16.31 | Noshei | ah darn Imhothar left |
| 01:17.32 | Noshei | was gonna tell him that RemoveExtraSpaces is kinda useless at any rate. It takes 5 consecutive spaces and removed 4 of them |
| 01:17.46 | Noshei | if any addon relies on that, wtf |
| 03:15.58 | seebs | ... what surreally strange code |
| 03:16.04 | seebs | the removeextraspaces thing. |
| 03:16.35 | seebs | That behavior sounds intentional, I think. They said it's run on chat messages. So it's probably supposed to turn arbitrary hunks of spaces into single spaces, but not remove the remaining space. |
| 03:16.43 | seebs | There is some crazy stuff going on in that UI. |
| 03:17.05 | seebs | For reasons I never figured out, back when I played, any chat message which contained a backslash followed by an n would instantly disconnect you from the server. |
| 03:17.24 | seebs | Note, there's no backslash escaping going on in any other way that I could find, so it's not as though this would have mattered, but there it is. |
| 06:35.22 | *** join/#riftuidev borbra (~borbra@grm-128-39-146-200.studby.uia.no) |
| 10:45.26 | *** join/#riftuidev Imhothar (~Imhothar@ppp-93-104-124-125.dynamic.mnet-online.de) |
| 12:08.10 | *** join/#riftuidev Noshei (~quassel@c-24-9-171-99.hsd1.co.comcast.net) |
| 18:09.25 | *** join/#riftuidev Imhothar (~Imhothar@ppp-93-104-124-125.dynamic.mnet-online.de) |
| 21:46.23 | *** join/#riftuidev Thorarin (thorarin@D57DD135.static.ziggozakelijk.nl) |
| 23:32.57 | *** join/#riftuidev borbra (~borbra@grm-128-39-146-200.studby.uia.no) |