IRC log for #bzflag on 20180621

00:21.18*** join/#bzflag infobot (ibot@208.53.50.136)
00:21.18*** topic/#bzflag is https://BZFlag.org || http://ohloh.net/p/bzflag || https://bzflag.org/help || Channel Logs: http://infobot.rikers.org/%23bzflag/ || 2.4.14 released! https://forums.bzflag.org/viewtopic.php?f=62&t=19953
00:21.18*** mode/#bzflag [+v infobot] by ChanServ
01:27.43*** join/#bzflag BZnotify (BZnotify@gateway/service/github.com/x-rewzjzcekpokoecu)
01:27.43BZnotify[13bzflag] 15blast007 pushed 1 new commit to 06master: 02https://git.io/fCVPI
01:27.43BZnotify13bzflag/06master 14189d7e0 15Scott Wichser: Don't pass an unnamed temp.
01:27.43*** part/#bzflag BZnotify (BZnotify@gateway/service/github.com/x-rewzjzcekpokoecu)
01:30.03JeffMThe c_str is const just send that in
01:30.17JeffMThen case to a const char*
01:47.08blast007for which part?
01:49.41JeffMin the callbacks, if you do't need it to be a std::string, you can send in just the c_str
05:32.20*** join/#bzflag FastLizard4 (fastlizard@wikipedia/pdpc.active.FastLizard4)
09:57.22*** join/#bzflag catay (~smertens@valyria.catay.be)
15:24.33*** join/#bzflag FusionDude (~flash@c-71-205-162-253.hsd1.co.comcast.net)
18:37.28*** join/#bzflag BZnotify (BZnotify@gateway/service/github.com/x-jlicbnrsjwdayvqe)
18:37.28BZnotify[13bzflag] 15JeffM2501 pushed 8 new commits to 06master: 02https://git.io/flFxL
18:37.28BZnotify13bzflag/06master 14c0c6434 15Jeffery Myers: move server side players into there own file...
18:37.28BZnotify13bzflag/06master 146e9b6ae 15Jeffrey Myers: make basic world object API stubs.
18:37.28BZnotify13bzflag/06master 14798aa26 15Jeffery Myers: move server side players into there own file...
18:37.28*** part/#bzflag BZnotify (BZnotify@gateway/service/github.com/x-jlicbnrsjwdayvqe)
18:52.29JeffM<PROTECTED>
20:06.48blast007bzfsAPIWorld.cxx:79:75: error: cast from type ‘const Obstacle*’ to type ‘void*’ casts away qualifiers [-Werror=cast-qual]
20:12.45blast007I was messing around passing a std::pair<std::string, void(*)(std::string)> to the libpng error callback, where the first part of the pair is the filename/filepath, and the second part is an alternative callback.  if the second is null, it'll just use std::cerr to write the error message but otherwise it will pass the message to the alternative callback
20:14.26blast007I set up the alternative function to use printError so it shows up on the panel instead of stderr, but only if the client is built debug or has debug level 1 or higher
20:19.37JeffMguess i'll need to keep a const void* in the pimple
20:25.10*** join/#bzflag BZnotify (BZnotify@gateway/service/github.com/x-kctnwawefsukmnmr)
20:25.10BZnotify[13bzflag] 15JeffM2501 pushed 1 new commit to 06master: 02https://git.io/flbIX
20:25.10BZnotify13bzflag/06master 14863a2f8 15Jeffery Myers: store a const ptr to the obstacle to make compiler happy.
20:25.10*** part/#bzflag BZnotify (BZnotify@gateway/service/github.com/x-kctnwawefsukmnmr)
20:25.16JeffMblast007: try that
20:25.49JeffMgcc really doesn't want you hiding data :)
20:26.41JeffMif that doens't work I'll have to build an obstacle ptr cache
20:30.14blast007if we'd still want to look at SQLite3 for our ban databases, there are some C++ wrappers that might be less terrible to work with: http://soci.sourceforge.net/doc.html  https://github.com/vinzenz/vsqlite--  https://github.com/SRombauts/SQLiteCpp    (the first two also use Boost, the third only depends on sqlite3)
20:31.58blast007there's also this one, but it's C++14 so that might exceed the support of our target platforms: https://github.com/SqliteModernCpp/sqlite_modern_cpp
20:33.08blast007SOCI also supports multiple database types, including Postgresql and MySQL
20:44.30JeffMI only have experience with the .net one
20:45.04JeffMand it's the same API for all database types since C# has a core network object
21:13.04JeffMyou could go with something like CDO, that abastracts the database
21:13.12JeffMit's like PDO for ++
21:25.11*** join/#bzflag Zehra (~Zehra@unaffiliated/zehra)
21:38.45JeffMblast007: I am testing a change that has windows use SDL2 for it's timer
21:44.49JeffMoh no.. that doesn't work at all :)
21:46.37JeffMahh it would help if I actualy divided by the QPF
21:47.28JeffMit was processing days of time every second
21:47.38JeffMthe background just flashed as it went day to night
21:48.45JeffMthe menu didn't work bevause it was was weeks between key presses :)
21:56.45blast007lol
21:57.24*** join/#bzflag spldart (~james@bzflag/contributor/spldart)
21:57.25*** mode/#bzflag [+v spldart] by ChanServ
21:59.05JeffMI'm not sure how the current "recalibration" code ever worked?
21:59.16JeffMit never actualy gets the new fequency
21:59.21JeffMit tries to compute it
21:59.34JeffMthat's never going to be right
22:03.22blast007yeah, it was pretty consistently wrong at lower cycles/sec
22:05.42*** join/#bzflag BZnotify (BZnotify@gateway/service/github.com/x-fhuuecccffzcwxaj)
22:05.42BZnotify[13bzflag] 15JeffM2501 pushed 1 new commit to 06master: 02https://git.io/flbn3
22:05.42BZnotify13bzflag/06master 149f6b94e 15Jeffery Myers: make windows use SDL's performance counter instead of WIMM and do a better job looking for downclocks....
22:05.42*** part/#bzflag BZnotify (BZnotify@gateway/service/github.com/x-fhuuecccffzcwxaj)
22:05.44JeffMgive that a shot. it should work better.
22:06.03JeffMif there are still issues I have an idea for how to resync
22:06.23JeffMI think the problem was the old code would detect freq changes when there were no
22:06.34JeffMI logged out the QPF and it was consitent
22:06.47blast007I think the problem only occured on super old dual-core AMD athon chips
22:07.02blast007and probably only on Windows 2000 or XP
22:07.03JeffMyet when I run the bzfs code, jsut at idle it's resyncing all the time.
22:07.20JeffMnow with this code I got no resync messages
22:07.28JeffMso try it out and let me know
22:07.42JeffMreally there should only be rare events when the CPU throttles up or down.
22:07.47JeffMlike going to battery
22:08.04JeffMif that's an issue we can use GetTicks to resync to the nearrest 10ms
22:08.40JeffMwith that change, it does mean that bzfs and bzadmin now require SDL2.dll
22:08.48JeffMon windows
22:08.58JeffMnot that it's an issue since you have them all in the same folder.
22:09.15blast007yeah, that should be fine for Windows
22:09.40JeffMI did not modify timekeeper for the other OSs
22:09.56blast007they were doing fine anyway, afaik
22:10.04JeffMyeah
22:10.14JeffMand I didn't want to have to mess with the link settings in make :)
22:11.03JeffMI bet this is why in my testing a windows server ends it's shots a little too early
22:11.03blast007it should be easier there than Windows.. I think it's just a $(LIBSDL) \ added to a Makefile.am
22:11.25blast007but in any case, shouldn't need it there
22:11.31JeffMit was pretty easy in windows, just add SDL2.lib to the library inputs
22:11.45JeffMhopefuly that works out better
22:16.42blast007for bzfsAPIServerSidePlayers.h, should plugins be #include'ing that, or should bzfsAPI.h pull that in?
22:16.58JeffMI'd say plugins should add it
22:17.04blast007k
22:17.08JeffMreally plugins should only pull in what they need
22:17.10JeffMto reduce bloat
22:17.27JeffMI'd love to have like bzfsAPITypes.h , bzfsAPIEvents.h etc...
22:17.37JeffMthen just have bzfsAPI.h include some of those for legacy plugins.
22:17.50JeffMmakes it easier to maintain IMHO
22:18.33*** join/#bzflag BZnotify (BZnotify@gateway/service/github.com/x-tbgykxadxvdupdud)
22:18.33BZnotify[13bzflag] 15blast007 pushed 1 new commit to 06master: 02https://git.io/flbcI
22:18.33BZnotify13bzflag/06master 14dd49ee0 15Scott Wichser: Include bzfsAPIServerSidePlayers.h in plugins that need it.
22:18.33*** part/#bzflag BZnotify (BZnotify@gateway/service/github.com/x-tbgykxadxvdupdud)
22:18.53JeffMthere were plugins that needed it?
22:19.20blast007yeah, two of them.  bzfscron and serverSidePlayerSample.
22:20.00blast007I wonder if they're not part of the fullbuild or bzfs solution
22:20.09JeffMthat's what I'm checking now
22:20.19JeffMcus I know I ran fullbuild at one point
22:20.27JeffMthat's usualy my last step before commiting
22:21.01JeffMthe sample for sure isn't
22:21.04JeffMbut cron is
22:21.15JeffMand yep it doesn't build
22:21.44JeffMwell 💩
22:31.47*** join/#bzflag BZnotify (BZnotify@gateway/service/github.com/x-nmqlkeopzgxdrnhg)
22:31.48BZnotify[13bzflag] 15JeffM2501 pushed 1 new commit to 06master: 02https://git.io/flbCL
22:31.48BZnotify13bzflag/06master 14673415f 15Jeffery Myers: whitespace... so much whitespace...
22:31.48*** part/#bzflag BZnotify (BZnotify@gateway/service/github.com/x-nmqlkeopzgxdrnhg)
22:40.37JeffMhmmm now I see why endshot is seperate from killed
22:40.43JeffMnot all kills end the shot
22:41.17JeffMbut it really should not be the client that decides what does and does not end the shot
22:42.06blast007it kinda has to be right now, though
22:42.16JeffMmaybe
22:42.18JeffMI am pondering
22:42.45JeffMmaybe I'll change it to 2 messages, MsgGotHit, and MsgKilled
22:43.02JeffMand with MsgGotHit the client will tell the server if it blowed itself up or not
22:43.35JeffMthen MsgKilled can have a bit for "end this shot" or not
22:43.38JeffMfor the remote players
22:44.01JeffMthen when there is more authoritive server, GotHit still makes sense.
22:44.53blast007will that break solo bots?  not that I care in the long run, of course
22:45.15JeffMno, the message is effectivly the same
22:45.30JeffMthe shotID will already be destroyed on the bothost client
22:45.43JeffMa benifit of global shot IDs :)
22:46.04JeffMthe only reason I'd split it is because I don't like messages that have different packings for server and client
22:46.24blast007so local player would still terminate shots that it gets hit with, but the server will decide if the remote players should see that shot terminate?
22:46.32JeffMthe client would still decide to end the shot or not, but only for it's local state, not everyone else.
22:46.36JeffMyes
22:46.57blast007do solo bots use the local player's state or do they have their own?
22:46.59JeffMit "should" not get out of sync with legit clients.
22:47.11JeffMthere is one state per client, used by N player objects.
22:47.33JeffMthere is one player list, one shot list, etc...
22:48.01JeffMwith solobots some of those players are "RobotPlayer" objects instead of "RemotePlayer" objects.
22:48.11JeffMso it should be ok
22:48.15blast007k
22:48.22JeffMthere is also only a single message processor on the client
22:48.40JeffMfor messages that can go to bots, they can "find local player" instead of "MyTank"
22:48.47JeffMso like chat is always MyTank
22:49.15JeffMbut death, flag drop etc calls FindLocalPlayer to see if it's the meat player or the robot players
22:49.26JeffMand if not then it looks for the remote players.
22:50.34JeffMMsgShotBegin for instnace first checks for local players to move the local only shots into the global shot list. That's how solo bots get there shot IDs
22:51.25JeffMcould have the flag send a "ends on hit" bit with the FlagType...

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