IRC log for #bzflag on 20171228

00:53.41*** join/#bzflag nadir (uid134094@gateway/web/irccloud.com/x-rykcvulzncljagxf)
01:18.38*** join/#bzflag infobot (ibot@rikers.org)
01:18.38*** topic/#bzflag is https://BZFlag.org || http://ohloh.net/p/bzflag || https://bzflag.org/help || Channel Logs: http://infobot.rikers.org/%23bzflag/ || 2.4.12 released! https://forums.bzflag.org/viewtopic.php?f=62&t=19720
01:18.38*** mode/#bzflag [+v infobot] by ChanServ
02:20.51*** join/#bzflag nitroxis (n@nxs.re)
02:32.07*** join/#bzflag nitroxis (n@nxs.re)
06:58.16*** join/#bzflag nadir (uid134094@gateway/web/irccloud.com/x-kpwqwercrnkpsewl)
13:51.47*** join/#bzflag nadir (uid134094@gateway/web/irccloud.com/x-sgzppvzpdwvniiid)
18:28.58*** join/#bzflag [Gort] (~gort@82-69-126-225.dsl.in-addr.zen.co.uk)
18:28.58*** join/#bzflag [Gort] (~gort@unaffiliated/gort/x-9231432)
18:33.03*** part/#bzflag BulletCatcher (~bc@bzflag/developer/BulletCatcher)
18:41.56*** join/#bzflag YepYep (~Mutter@12.130.118.61)
19:15.23JeffMso is there any actual desire/interest in another bzfs server ?
19:20.23blast007meaning an alternative implementation?
19:20.35JeffMyes
19:20.40blast007I don't know
19:21.01JeffMI've got it to the point where it's processing all the messages for regular play
19:21.49JeffMwell for a non flag server :)
19:22.05JeffMbut it's tracking all the flags and stuff, so it has a place to do the flag messages.
19:22.20blast007so no CTF yet either?
19:22.30JeffMno, no flag caps
19:22.34blast007k
19:22.41JeffMbut flags are maybe a couple hours of work
19:22.54JeffMit's just filling in all the bits
19:23.13JeffMand writing modules to handle all the config features that would be needed.
19:23.34JeffMlike planting random flags and stuff
19:23.42blast007I could run an instance or two alongside my normal bzfs instances if you're wanting a long-running public test
19:24.06JeffMI think a short term public test would be more warranted :)
19:24.16JeffMor even short term non-public tests
19:24.21JeffMI've only ever run it localhost
19:25.44JeffMI want to do some more testing with shots, but the rest of the work is tedious if there is no interest
19:30.51blast007is it written in C#?
19:31.44JeffMyes
19:32.29JeffMseems to build fine under monodev on ubuntu, so I'd expect it to build fine under msbuild on any linux that has a mono that supports .net 4.6.1 or later
19:32.54JeffMbut then you also don't need to build your target OS if you don't want to
19:33.15blast007In my opinion, that'd be most useful if we were also moving the client to C#.  Otherwise there's going to be duplication of code/effort keeping things in sync.  The threading aspect is interesting though.
19:34.21JeffMthe netlib wold be the the main thing that would get duplicated, but it's a prety easy translation and it only changes when proto changes.
19:34.22blast007not maybe as much duplication now, but there'd be more if we started to move to a proper server-side state, since client and server would need to have the same simulation running.
19:34.33JeffMyes,
19:35.25JeffMthe C# components could be used for a unity based client.
19:35.36JeffMor one using normal openGL
19:36.52JeffMit's probalby the only viable language where there will be no fights over different build systems :)
19:36.58JeffMbecause there is only one.
19:38.25blast007the fights were more over build system generators ;)  (as in, cmake and premake would be making the visual studio projects or makefiles)
19:38.34JeffMexactly
19:38.48JeffMthere is only one file that all systems make, the .sln file
19:38.53JeffMthere are no make files
19:39.01JeffMall OSs use the same files.
19:39.09blast007you could still using cmake or premake
19:39.11JeffMso there is no need for a seperate layer to manage them
19:39.13blast007use*
19:39.16JeffMsure but it'd be silly
19:39.29blast007it might be necessary for the linuxes
19:39.33JeffMnope
19:39.36blast007they'd want to use system libs
19:39.43JeffMit's not native code, so no
19:39.51JeffMnuget handles deps
19:40.02JeffMnuget restore BZFSPro.sln
19:40.13blast007talking about packaging for linux, not our own bin distribution
19:40.15JeffMmsbuild BZFSPro.sln
19:40.32JeffMagain, it's not native code, it does not include libs or what you thihk of as "bins"
19:40.44JeffMthe same physical EXE runs on every single platform.
19:40.52JeffMso there is no packaging step like you think of
19:40.56blast007but there's dlls, no?
19:41.09JeffMthey are all built by the project or handled by nuget
19:41.21blast007yeah, that's what Debian doesn't like :)
19:41.36JeffMthe ones that debian manages are part of mono
19:41.47JeffMand provided by debians mono install
19:41.54JeffMso they have nothing to complain about.
19:42.15JeffMright now it only uses a single nuget package, the YAML thing
19:42.44JeffMdebian ships several mono packages.
19:42.50JeffMso does ubuntu
19:43.05JeffMyou'd just have to make sure the right mono was installed.
19:43.15dngorI think the build system tail is wagging the rewrite it in a different language dog.
19:43.35JeffMno, the build system was not why I did this
19:43.50dngorIt's how you're selling it right now.
19:43.56JeffMI'm just saying that it's diferent enough to make the previous discussions moot.
19:44.08JeffMyou don't know the history, I'm just mentioning it
19:44.19JeffMthe main thing is it's better language and a better codebase
19:44.27JeffMhe's aware of those aspects.
19:44.55JeffMbut for sure there would be changes to packaging
19:45.20JeffMmain selling point, it doesn't support solo bots :)
19:48.02blast007I do wonder if it would needlessly complicate some other aspects though, like selecting other third-party libs.  Like, let's say we want to use OpenAL Soft.  There is at least one .NET binding for OpenAL, but is that using a native code version of OpenAL?  I'm not real familiar with how all that works.
19:49.02JeffMyeah Open TK has bindings to OpenAL, so it'd use the platforms native openAL and call it from a wraper.
19:49.11JeffMyou can write wrapers to native C++ libs.
19:49.16JeffMpnivoke and marshaling.
19:49.36JeffMso you'd package with a dependency to openAL as normal, and include your wraper code.
19:49.56JeffMyou don't need special builds of openAL or anything
19:50.16blast007ok
19:50.28JeffMthe mono runtime efectivly calls loadLibary etc...
19:51.09JeffMthere are even tools to generate wrapers automaticly if you need.
19:55.39JeffMcode wise, I belive it's much cleaner, here is an example of a network message,  https://www.irccloud.com/pastebin/b9MB7WYY/
19:56.09JeffMthe game code only deals with message objects, so there isn't pack/unpack stuff spread all over.
19:56.45blast007pretty sure we could enter some of our existing code to The Internation Obfuscated C Code Contest
19:57.06JeffMis there an "unintentional" category?
19:57.28blast007sorry, International*
19:58.00blast007dunno about categories :)
20:01.02*** join/#bzflag BulletCatcher (~bc@bzflag/developer/BulletCatcher)
20:01.02*** mode/#bzflag [+v BulletCatcher] by ChanServ
21:01.06JeffManother good thing is it's total lack of support for /password :)
21:01.19blast007heh
21:18.18blast007JeffM: can't remember if you had mentioned this or not - how challenging would it be to make yours work with IPv6?
21:18.50JeffMshould work right now
21:19.08blast007nice
21:19.27JeffMit's using the .net TCPClient class, that abstracts the raw socket
21:19.39JeffMand I belive I ask for it listen on anything
21:20.16JeffMworst case scenario it would mean adding a second TCPConnectionManager instance with a different listen settings
21:20.27JeffMbut they'd still be able to feed into the same game system
21:20.39JeffMthe server is designed to allow multiple ports to host the same game.
21:21.58JeffMand maybe the banlist processing
21:23.14blast007I had a start to a new banlist class in C++, though I think it used some linux-specific stuff and also didn't use the proper data types
21:23.39JeffMthe structure of this is a little different.
21:23.54JeffMthe main applicaiton is more of a framework for stuff
21:24.14JeffMit realy only handles things that are direclty part of the game protocoll
21:24.33JeffMthe actual server management, and "game" stuff is handled by various managers and modules.
21:25.00JeffMso the server itself doens't know about "permissions"
21:25.13JeffMthat's all handled by a module
21:25.41JeffMso that all the security related stuff can be maintained as it's own thing, not mixed into larger proejct
21:25.58JeffMthe bad word filter is it's own DLL, and called by the chat processing thread for example.
21:26.08JeffMI'll do /commands the same way
21:30.34JeffMahh, looks like I'm only listening on V4 interfaces
21:30.38JeffMI'll add a V6 listener
21:31.31JeffMblast007: you'd always want to listen on both types on the same port right?
21:35.26JeffMthere we go, IPV6 support
22:10.23blast007yeah, it'd be the same port
22:10.54JeffMthen it's done, just added another listener and shunted connectons into the existing pocessor
22:13.41blast007I'll have to change global auth to work with IPv6 as well
22:13.48JeffMsure
22:13.54blast007make the token tied to a specific server instead of a player IP
22:13.59JeffMand, you know.. the clients :)
22:16.42blast007both changes should be reasonably easy
22:24.57JeffMforgot UDP
22:25.36JeffMand done
22:25.56JeffMwould the game ever want to know if a user was IPV4 or 6?
22:26.05JeffMother than ban look ups of course
22:30.24blast007currently we have a /playerlist command that shows the IP
22:30.34JeffMwhy?
22:30.42blast007and admins are sent the IP addresses of players
22:30.48JeffMwhy?
22:31.08blast007to assist with bans
22:31.23JeffMwhy enter bans in Ips?
22:31.29blast007/playerlist also shows their hostname
22:31.49JeffMwell still, this has the address as a string for display/logging so that's not an issue
22:31.52blast007right, "currently"
22:32.01JeffMI mean does the player ojbect need a flag "IAmIPV6"
22:32.02blast007it's not scalable to IPv6
22:32.29JeffMI was going to just do bans by user, and options to do IP, or subnet
22:32.29blast007I don't know
22:32.55JeffMwell it's a property of the RemoteEndpoint strucutre, so it could be gotten
22:33.18JeffMmaking people type in IPs is silly :)
22:33.43JeffMlike /ban player [BYID/BYIP/BYHOST/BYSUBNET]
22:33.59JeffMand the server can figure out what the rest meaans
22:34.33JeffMor may be just /ban player AUTO, and have it do ID if they have an ID, and IP, and let the server have an option if it wants to alos apply host and subnet.
22:35.23JeffMI was thinking of storing bans in a SQLite db, where each record could have one ban of each kind.
22:35.27JeffMso they'd be linked
22:40.05blast007I'd like to eliminate the need for the vast majority of the slash commands
22:57.38JeffMWhat ones are needed in your opinion?
22:58.25blast007well, /me  :)
22:59.06JeffMYou want kid/ban to have there own messages?
22:59.16blast007possibly /quit, /part, and /msg commands, possibly others that are commonly used in hotkeys
22:59.27blast007kick/ban can be part of a dialog
22:59.49blast007so, it'd be part of the messages for handling dialogs
22:59.55JeffMSo just shunted to another MsgType
23:00.50JeffMWhat would request the dialog from the server?
23:01.04JeffMOr would the client initiate the dialog?
23:03.36blast007client would initiate the request for the dialog
23:04.22JeffMso there would be some sort of "ban" function built into the client that would start the process
23:04.58JeffMmy intent is to implment kick and ban functions as features of the code, but not tie them to any specific triggering mechanism
23:05.14JeffMAPI could call them, /command could call them, or watever, dialog processor could call them too
23:20.23JeffMhow does mapchange force the world reload?
23:20.27JeffMdoes it boot everyone?
23:35.37JeffMoh I guess it restarts the exe.. lame
23:44.57*** join/#bzflag usumbucu11 (uid271915@gateway/web/irccloud.com/x-nibnjvdywetgdbgl)
23:47.40JeffMallejo , macsforme : based on what I have learned of the game protocoll, you could kill everyone, temporarly disable spawns, and send MsgGameSettings to all clients and that should force them to reload the map.
23:48.54JeffMthat should force the client back into a downloadworld loop
23:49.39JeffMmay not hurt to resend BZDB vars too
23:50.11allejowouldn't players still need to rejoin?
23:50.21JeffMnope, that won't kill the connection
23:50.24JeffMdon't use superkill
23:50.39JeffMlike tell them they are dead :)
23:50.48JeffMthe connection stays open
23:51.07JeffMit just puts you back to the start of the connect sequence.
23:51.17JeffMthe client should continue like it would on nomral startup
23:51.34JeffMit'll re-request the world, and send it to a new WorldBuilder
23:51.41JeffMand that will reinit the sceenenodes.
23:51.54JeffMand then they'll go back to request the inital spawn
23:52.11JeffMbut the player list will already be filled out
23:52.24JeffMether that or it'll set everything on fire
23:53.09allejowould this be at all similar to the current replay functionality? where it'll load the new world but still require a client rejoin due to a bug, or so server messages say
23:53.48JeffMI'm not aware of that part
23:53.58blast007sounds like you could fix the replay servers so they don't require a rejoin
23:54.16JeffMI just know I was able to get the client to request the world twice when I had some messages screwed up
23:54.37JeffMthe client's handling of MsgGameSettings is what triggers that process
23:55.22allejoyour own client or bz's client? cuz i'm open to testing it out to see if it works. bzfs would need to load the new world file too though, no?
23:55.28JeffMthe real BZ client
23:55.39JeffMI was connecting it to my server
23:56.23JeffMI had a buffer problem  and was sending the settings twice, once over TCP and once over UDP
23:56.34JeffMI was wondering why the client would load the map twice
23:56.51JeffMbut the important thing for you is, the client actualy loaded the map twice.
23:57.19JeffMif it can do it twice, it can do N time :)
23:57.26allejobzfs currently has a partial API event, bz_eGetWorldEvent which will reload the world file, however it won't parse the options block. bzfs will still need to reread the new world file and parse it correctly before it resends it to the client, right?
23:57.34allejohttps://github.com/BZFlag-Dev/bzflag/issues/5
23:57.47JeffMsure, but you can FIX bzfs.
23:57.52JeffMyou can't fix clients in the field.
23:58.05allejothat's fair
23:59.39JeffMthat would be the way I'd implment it in my server

Generated by irclog2html.pl Modified by Tim Riker to work with infobot.