| 00:47.15 | *** join/#bzflag alezakos (~kongr45gp@95.46.199.28) |
| 01:12.32 | *** join/#bzflag alezakos (~kongr45gp@95.46.199.28) |
| 04:35.05 | *** join/#bzflag FastLizard4|away (fastlizard@wikipedia/pdpc.active.FastLizard4) |
| 04:52.48 | *** join/#bzflag zuii (~akos@BC06BF93.catv.pool.telekom.hu) |
| 05:40.12 | *** join/#bzflag FastLiza1d4|away (fastlizard@wikipedia/pdpc.active.FastLizard4) |
| 06:32.20 | *** join/#bzflag I_Died_Once (~I_Died_On@unaffiliated/idiedonce/x-1828535) |
| 06:58.20 | *** join/#bzflag Shuist (~Shuist@ppp203-122-213-220.static.internode.on.net) |
| 12:21.13 | *** join/#bzflag alezakos (~kongr45gp@bzflag/developer/alezakos) |
| 14:17.41 | Constitution | if c-ares is part of the problem, should we look again at replacing it with gethostbyname() in a thread? would be awesome to eliminate that dependency |
| 14:18.48 | blast007 | but then we wouldn't get free publicity from the c-ares website :P |
| 14:19.16 | *** mode/#bzflag [-o blast007] by blast007 |
| 14:20.35 | Constitution | yes, we'll lose all of our 0.5 new players per month :-P |
| 15:09.09 | BulletCatcher | Our use of threads for screenshots seems to be reliable, so the same approach ought to work for asynchronous DNS lookups. |
| 15:09.15 | BulletCatcher | Any platforms without POSIX threads could still use c-ares. |
| 15:09.17 | BulletCatcher | Note that the Linux man page says gethostbyname(3) is deprecated in favor of getaddrinfo(3), which is thread safe. |
| 15:36.52 | blast007 | I don't believe we're having any issues with c-ares on platforms other than Windows |
| 15:37.27 | blast007 | (though, iirc, some of the more obscure platforms may lack c-ares in their package managers) |
| 15:49.32 | *** join/#bzflag JBdiGriz (~dbw@108-197-33-160.lightspeed.sndgca.sbcglobal.net) |
| 15:54.04 | blast007 | looks like curl itself can use either c-ares or a threaded implementation for async DNS, so should be no problem not using it on Windows. |
| 15:54.42 | blast007 | Honestly, if we can get by with just not linking libcurl with c-ares on Windows for the 2.4.4 release (and leaving the c-ares for our other DNS resolutions), I think that's good enough |
| 15:55.08 | blast007 | We can then look into eliminating it more completely for 2.5/2.6. |
| 15:57.11 | blast007 | I build a new installer a day or two ago that doesn't like c-ares to libcurl and gave it to Mopar Madness to test out. He was one that was definitely seeing some crazy long delays for HTTP requests. Just waiting to hear back. |
| 15:59.56 | *** join/#bzflag zuii (~akos@BC06BF93.catv.pool.telekom.hu) |
| 16:02.07 | blast007 | https://download.bzflag.org/2.4.4/windows/bzflag-2.4.4.20151020-RC1-alternate.exe <-- that's the installer, in case anyone else was having delayed MOTD, server list, and authentication on the Windows RC1 client. |
| 16:03.22 | blast007 | be curious to hear if that takes care of the problem. |
| 17:07.32 | *** join/#bzflag zuii (~akos@BC06BF93.catv.pool.telekom.hu) |
| 19:08.50 | *** join/#bzflag zuii (~akos@BC06BF93.catv.pool.telekom.hu) |
| 20:03.03 | *** join/#bzflag zuii (~akos@BC06BF93.catv.pool.telekom.hu) |
| 20:19.16 | *** join/#bzflag zuii (~akos@BC06BF93.catv.pool.telekom.hu) |