[GEN][ZH] Add compilation for Linux

[feliwir]

Based on #479
Targets fixed:

Generals:

• g_compress
• g_wwvegas
  • g_wwlib
  • g_wwdebug
  • g_wwmath
  • g_wwsaveload
  • g_wwutil
  • ....

Zero Hour:

• z_compress
• z_wwvegas
  • z_wwlib
  • z_wwdebug
  • z_wwmath
  • z_wwsaveload
  • z_wwutil
  • ...
• ...

[zzambers]

Is there a reason, why debug directory was renamed to Debug?

[zzambers]

I think some alternative implementation should be provided or some kind runtime error should be produced...

[feliwir]

I'll use GLM here (as we do on our fork). However first i want to integrate VCPKG (possibly a separate PR). For linux to work we need a few external libraries and VCPKG makes that a lot easier.

[xezon]

Not sure if this specific line is that, but what happens in the code at many places is a row major to column major matrix conversion or vice versa. We polished this in Thyme and it needs to be done in this code as well. This needs to be done in a separate change because it is bigger change. Here is the reference:

TheAssemblyArmada/Thyme@2114738

Comment from that change:

// When converting Matrix4 to D3DMATRIX or vice versa always use conversion function below.
 
// Reason being, D3DMATRIX is row major matrix, and Matrix4 is column major matrix.
 
// Thus copying from one to another will always require a transpose (or invert).

I have opened task with #552

[feliwir]

For reference this is the GLM implementation we have: https://github.com/Fighter19/CnC_Generals_Zero_Hour/blob/develop/GeneralsMD/Code/Libraries/Source/WWVegas/WWMath/matrix3d.cpp#L567

[zzambers]

@feliwir I have no doubts that you plan to implement it later. :) My point was more about approach to unimplemented code. I think there should be some unified way how to handle this, right from the start, otherwise there is a danger of silently broken code, if it gets forgotten. Also It would be nice to be able to search for unimplemented code. I have created issue, where I formulated this in more detail.

[zzambers]

I think some alternative implementation should be provided or some kind runtime error should be produced...

[feliwir]

We can use std::mutex here, however i'm not sure if the memlogging is used much? When searching for function invocations i only get a few, which don't need any locking mechanism (only ww3d.cpp)

[OmniBlade]

Why not just replace this version of wwmemlog with the version from ZH which seemingly is more linux friendly?

[feliwir]

I didn't look at that one yet, but when we do dedup first, this probably will resolve itself

[feliwir]

Is there a reason, why debug directory was renamed to Debug?

Yes, otherwise it comflicts with GCC standardlibrary, which includes a header named debug/debug.h

[zzambers]

Note, that this is related to my PR to make code compile with MinGW toolchain as GCC compiler is used in both cases. I think it would make most sense for this to go on top of MinGW support, because:

• MinGW uses windows-compatible headers, so are changes are mostly fixes for GCC. I have changes for ZH (game + internal libraries) basically done (most changes are in engine). Now I just want to redo commits (combining some, for easier review). Then I plan to do backport to GEN.
• Linux port on the other hand needs fixes for GCC + replacements for windows-specific stuff, so when finished is mostly superset of MinGW effort. So I think it would make sense to (re)base it on MinGW work.

btw. build fixes for GCC also slightly overlap with VC warning fixes (+ need asm replacements).

[feliwir]

Will be redone after commonizing