| <<<Back 1 day (to 2016/03/09) | 20160310 |
Robin_Watts | Well, that's odd. | 10:31.08 |
| I can't build memento builds on windows. | 10:31.20 |
| even old versions. | 10:31.31 |
chrisl | Ah, kens noticed that in SF.... not sure if he got to the bottom of it | 10:33.02 |
kens | ? | 10:33.24 |
chrisl | memento not building on Windows | 10:33.48 |
kens | I never got it fixed, I just commented out windows.h and the code that used it | 10:34.05 |
| very hacky :( | 10:34.15 |
Robin_Watts | I'm utterly stumped. | 10:35.21 |
kens | me too | 10:35.40 |
Robin_Watts | even moving #include <windows.h> to the top of the file so it's the first thing that gets included, it fails during reading that file. | 10:35.49 |
| It's as if something has broken windows.h | 10:35.56 |
kens | yeah tried that too | 10:35.57 |
Robin_Watts | Like a windows update or something. | 10:36.05 |
kens | But we include windows.h in the GS windows build | 10:36.14 |
| and it works there | 10:36.26 |
Robin_Watts | yes. | 10:36.31 |
kens | So its not the file, or its not including the same file | 10:36.43 |
| I had intended to try thsi on my system here, in case it was the weird setup I have in my laptop, but I forgot | 10:37.17 |
chrisl | What about using windows_.h instead? | 10:37.17 |
Robin_Watts | still fails :( | 10:38.48 |
chrisl | I guess Windows is just sh*t..... can you tell I'm arguing with it, too?? | 10:39.47 |
kens | I'm sure it used to work, try a bisect ? | 10:40.14 |
Robin_Watts | kens: It certainly used to work. | 10:40.26 |
| My machine did a windows update last night... | 10:40.42 |
malc_ | WIN32_LEAN_AND_MEAN | 10:40.43 |
kens | well that is possible, but I was using an include in windows sdk | 10:40.56 |
Robin_Watts | kens: I've backtracked beyond any of my recent changes to memento. | 10:41.52 |
| malc_: Is that intended as a helpful suggestion, or just a random interjection? | 10:42.07 |
kens | maybe its a build change | 10:42.20 |
chrisl | WIN32_LEAN_AND_MEAN "reduces the size of the Win32 header files by excluding some of the less frequently used APIs" | 10:42.45 |
kens | Sadly I don't remember when I last built it and it worked | 10:42.45 |
Robin_Watts | ah, ok, so I have a point at which it worked. | 10:43.37 |
malc_ | Robin_Watts: both actually | 10:43.53 |
kens | OK so a bisect then ? | 10:44.33 |
Robin_Watts | It goes wrong with: b405260 | 10:44.49 |
malc_ | uhm... there's no such sha id here | 10:46.26 |
chrisl | +#include "mupdf/memento.h" is probably not good...... | 10:47.26 |
Robin_Watts | chrisl: indeed. | 10:47.34 |
| I fear it may go wrong there because it's the first time I include windows.h | 10:47.54 |
chrisl | Baffled about why it would cause problems with windows.h though | 10:47.54 |
Robin_Watts | That include is never taken. | 10:48.14 |
| I have a #define MEMENTO_GS_HACKS thing that bends includes to suit gs's weird environment. | 10:48.44 |
chrisl | Oh, I was misreading the #ifdefs - I wish the preprocessor was better with indenting.... | 10:49.54 |
| Okay, how do delete no longer available network drive mappings when Windows insists on trying to connect to them on any mouse click on them??? | 10:52.47 |
kens | cli | 10:53.03 |
| THere's a command line for it, but don't ask me what it is | 10:53.16 |
| net stop something maybe ? | 10:53.33 |
chrisl | Aha! "net use /del <drive letter>" | 10:54.42 |
| Thank you | 10:54.49 |
kens | NP | 10:55.19 |
chrisl | Hmph, returns an error :-( | 10:55.21 |
kens | LOL | 10:55.26 |
chrisl | It works for connected network drives..... <sigh> | 10:56.05 |
Robin_Watts | ok, so if I backtrack to the working memento.c and add #include <windows.h> to the top of it, it dies in the same way. | 10:57.26 |
kens | sadly not helpful really | 10:57.41 |
| do we have different compile flags for base and psi ? | 10:59.05 |
chrisl | Erm, probably.... | 10:59.42 |
kens | well memento is in base, and dwmain and friends are in psi I thnk | 10:59.56 |
| If the compile flags are differnt that might explain it, maybe... ANyone got a straw ? | 11:00.23 |
chrisl | windows.h is included in base as well | 11:00.24 |
kens | Then I give up | 11:00.34 |
Robin_Watts | So /Za disables microsoft extensions. | 11:00.40 |
| winnt.h requires microsoft extensions to be enabled. | 11:00.53 |
| cos anonymous structs are an MS extension. | 11:01.12 |
| So we can't use /Za when building memento. | 11:01.23 |
chrisl | Robin_Watts: that's true for LCMS2, as well. If you look in lcms2.mak it's got a compiler incantation you can base a memento one on | 11:02.39 |
Robin_Watts | chrisl: Fab, thanks. | 11:03.25 |
chrisl | The other such places uses GLCCWIN which wouldn't be good in a non-Windows specific makefile ;-) | 11:04.24 |
Robin_Watts | memento is not in a windows specific makefile. | 11:04.58 |
chrisl | Exactly my point.... and I'd rather not move it to platform specific ones | 11:05.33 |
Robin_Watts | gotcha. THanks. | 11:08.47 |
| chrisl: Third commit on robin/master is my proposed makefile fix. Care to cast your eye over that? | 11:28.00 |
chrisl | Sure..... | 11:28.20 |
| Robin_Watts: That looks spot on | 11:29.38 |
Robin_Watts | chrisl: Thanks. | 11:33.56 |
chrisl | I'm going to be out from now until late afternoon...... | 11:39.16 |
Robin_Watts | have fun. | 11:39.29 |
chrisl | As long as the traffic isn't too bad..... cheers! | 11:39.57 |
Robin_Watts | oh, ffs. | 11:55.30 |
kens | ?? | 11:55.37 |
Robin_Watts | Works on windows. not on linux. | 12:05.05 |
| I'm having a bad day. | 12:05.19 |
kens | O.O | 12:05.20 |
Robin_Watts | All the rest of the source is being built with -DMEMENTO, except memento itself. | 12:06.46 |
kens | that seems perverse | 12:07.23 |
Robin_Watts | You're telling me. | 12:07.54 |
| So, on unix, the flags get bundled into CC_ | 12:15.17 |
| whereas on windows they are in CFLAGS, as you'd expect. | 12:15.30 |
| OK, I think they are in GENOPT. I can use that to work around it. | 12:25.48 |
| tor8: 1 memento fix on robin/master then. | 12:54.43 |
| Trying to keep gs and mupdf in sync. | 12:54.49 |
tor8 | Robin_Watts: memento fixes LGTM | 14:08.08 |
| Robin_Watts: a whole bunch of commits on tor/master for review as well | 14:08.32 |
| I'm seeing something disturbing with the jni bindings | 14:08.42 |
| in ft_kind, FT_Get_Font_Format returns NULL and segfaults | 14:09.13 |
| if I change it back to FT_Get_X11_Font_Format it works | 14:09.22 |
| *but* ... the implementation of both functions are identical | 14:09.32 |
| and it only happens for java | 14:09.59 |
Robin_Watts | That is indeed odd. | 14:23.36 |
| Forward 1 day (to 2016/03/11)>>> | |