Closed chewi closed 2 years ago
I've realised this did break the Windows build, or at least Visual Studio 2022, due to it picking up the wrong parser.h
, but as far as I can tell, it was already broken anyway. Despite what my profile might suggest, I don't care about the Windows build that much, but with yours being one of the more prominent forks, it would be a shame to leave it broken. I think I've nearly got it working, so I'll see what I can do.
Ah yes, if the windows build can be fixed, that would be great. I haven't tried a windows build in ages, but I'll check how it works at some yes, it would be good. (also, I'll to work on some CI at some point)
I've been making some progress but it's very slow. It's more broken than I thought. When did you last do a Windows build and was it with Visual Studio or MSVC? At least Visual Studio seems broken since Ryan's changes. For example, it looks for #include "StdH.h"
in most source files to use the precompiled headers, but Ryan changed most of these to #include "Engine/StdH.h"
or similar so it breaks. I can fix that, but it's just the tip of the iceberg. Or maybe you'd prefer to drop Visual Studio and just use CMake? There seem to be a lot of other MSVC-specific issues that would affect CMake too though. I briefly tried MinGW, but I don't think that ever worked.
Can't remember, it was quite a long time ago. I have some VS2019 installed...
Stange. Maybe earlier versions didn't care if the precompiled headers weren't used then. Oh well, I'll persevere for now.
This makes cross-compiling easier, and saves having to build once each for TFE and TSE.
I considered using CMake's import/export feature, but there's little point when you're only importing one binary target. Pointing the ECC variable at the ecc binary itself is simpler.
I copied the minimum CMake version of 2.8.7 from the parent project. To be confident this would actually work, I tested building under CentOS 7 with CMake 2.8.12.