Log for on 11th August 2015:
Times are UTC Toggle Colours
00:50:13  *** keoz has quit IRC
01:25:48  *** Supercheese has quit IRC
02:04:19  *** Supercheese has joined
02:04:19  *** ChanServ sets mode: +v Supercheese
02:08:26  *** Supercheese has quit IRC
04:09:48  *** Supercheese has joined
04:09:48  *** ChanServ sets mode: +v Supercheese
11:32:51  *** Supercheese has quit IRC
11:33:22  *** Supercheese has joined
11:33:22  *** ChanServ sets mode: +v Supercheese
12:22:09  *** Alberth has joined
12:22:09  *** ChanServ sets mode: +v Alberth
16:18:11  *** frosch123 has joined
16:18:11  *** ChanServ sets mode: +v frosch123
17:18:29  <michi_cc> Rubidium: I build some libs (read icu) without debug info as to not blow up the zip too much, but I can rebuild them with debug info for the compile farm if you want.
17:18:57  <michi_cc> I'd still use the PDB-less libs for the general public download though.
17:19:36  <Rubidium> oh, it gets embedded into the .lib instead of a seperate PDB?
17:20:46  <Rubidium> but yes, I'd reckon a seperate one for the CF might not be that bad, except for reproducable builds
17:21:50  <Rubidium> but I'd reckon that we provide the extern libraries as courtesy, so license wise it shouldn't be a problem
17:24:13  <michi_cc> Rubidium: Both actually, the ICU libs are bigger with debug info, but you still need to supply the PDB as well.
17:25:40  <Rubidium> how much does it affect the final binary?
17:59:57  <michi_cc> Rubidium: It doesn't have any effect on the exe, debug info there is always separate. The PDB does grow though.
18:06:44  <Rubidium> PDB size isn't really a problem
18:06:46  <michi_cc> Rubidium:, and I've also overwritten with a new version as I just noticed the libpng I included previously would output a warning about missing debug info when building OTTD.
19:11:19  <Rubidium> I was going to spend the night on getting OpenTTD built with a new GCC on Windows...
19:11:24  <Rubidium> it doesn't compile yet though...
19:11:46  <Rubidium> apparantly SHGFP_TYPE_CURRENT is defined now (not sure how exactly yet)
19:26:07  <Rubidium> <- looks kinda dangerous, but I wouldn't know how to check for newness of MINGW in this case
19:26:38  <Rubidium> fios.h includings win32.h via some other headers
19:28:59  <frosch123> #if (_WIN32_IE >= 0x0500) <- google suggests that
19:34:35  <Rubidium> /src/stdafx.h:         #define _WIN32_IE_ 0x0600         // 6.0 (XP+)
19:40:11  <frosch123> hmm, so you want to check for some older version of mingw32 header files
19:40:25  <frosch123> how does _SHLOBJ_H_ help?
19:40:38  <frosch123> wouldn't that one be defined whenever the headerfile exists?
19:41:32  <Rubidium> yes, and when it doesn't exist... it should do the define
19:42:44  <frosch123> doesn't gcc5 exit if a header file is missing?
19:43:30  <Rubidium> hmm... good point-ish
19:43:43  <Rubidium> this modification will ofcourse fail with older mingws
19:52:33  <Rubidium> the default libraries don't want to play nice :(
19:53:20  <Rubidium> when linking statically some libraries aren't mentioned causing link failures
19:54:22  <Rubidium> mostly it's freetype that includes harfbuzz that needs other libraries that aren't mentioned in pkg-config --libs
19:55:09  <Rubidium> not that freetype-config is any better
19:55:14  <Rubidium> ls
20:11:41  *** Alberth has left
20:13:13  <Rubidium> ah well, tomorrow (or after that) I might try again
20:13:39  <Rubidium> and if it gets too complex, then... I'll not add/update the MinGW manual and we'll just use the botched version from here ;)
20:24:01  <LordAro> on that subject, i've been meaning to look into msys2, which apparently makea things much easier
21:37:09  *** frosch123 has quit IRC
23:29:56  *** Supercheese has quit IRC
23:29:56  *** gnu_jj_ has quit IRC
23:29:56  *** tycoondemon has quit IRC
23:29:56  *** Rubidium has quit IRC
23:31:20  *** Supercheese has joined
23:31:20  *** gnu_jj_ has joined
23:31:20  *** tycoondemon has joined
23:31:20  *** Rubidium has joined
23:31:20  *** sets mode: +vv Supercheese Rubidium

Powered by YARRSTE version: svn-trunk