I agree with the idea. I have a bit of concern about execution.

diff --git a/CMakeLists.txt b/CMakeLists.txt
index ad57a86b5..da03f6d83 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -530,7 +530,7 @@ if(WIN32 AND MSVC)
     endif()
 endif()
 
-if(NOT WIN32 AND NOT APPLE)
+if((NOT WIN32 AND NOT APPLE) OR MINGW)
     install(TARGETS ${PROJECT_NAME}
         RUNTIME DESTINATION bin
         LIBRARY DESTINATION lib

The logic of change to line 533 leaves possibility of Unix, Win64, Linux, et al.
I'm unsure how best to fix this but I'm leery of the change suggested.

The logic of change to line 533 leaves possibility of Unix, Win64, Linux, et al.

Doesn't previous logic also include those possibility? There is no Win64. WIN32 includes Win64. ¹

There is another logic that can be used. Does this look good?

@@ -530,7 +530,7 @@ if(WIN32 AND MSVC)
     endif()
 endif()

-if(NOT WIN32 AND NOT APPLE)
+if(NOT MSVC AND NOT APPLE)
     install(TARGETS ${PROJECT_NAME}
         RUNTIME DESTINATION bin
         LIBRARY DESTINATION lib

The logic of change to line 533 leaves possibility of Unix, Win64, Linux, et al.

Doesn't previous logic also include those possibility? There is no Win64. WIN32 includes Win64. 1

There is another logic that can be used. Does this look good?

@@ -530,7 +530,7 @@ if(WIN32 AND MSVC)
     endif()
 endif()

-if(NOT WIN32 AND NOT APPLE)
+if(NOT MSVC AND NOT APPLE)
     install(TARGETS ${PROJECT_NAME}
         RUNTIME DESTINATION bin
         LIBRARY DESTINATION lib

Footnotes

1. https://cmake.org/cmake/help/latest/variable/WIN32.html [leftwards_arrow_with_hook](#user-content-fnref-1-00f14395ee875d1af3c14bc972778015)

A little further up in CMake docs is Variables That Describe the System.

MSVC is implying a MS Visual Studio Build. This may imply a Windows environment but is not exclusively a Windows system, IIRC.

I think what may be best is that rather than an exclusive condition coupled with an inclusive condition (i.e. (NOT a and NOT b) or C) the logic test should just be simply exclusive only (i.e. NOT a and NOT b and NOT...) or inclusive only (i.e. a or b or c). Mixing seems bad to me, IMHO.

My apology, I am out of ideas 😔

Hmmm, is this a situation where we are potentially rejecting a reasonable PR because the existing approach is too wide open?

Looking at the suggested change, it seems to only add MINGW to the list of environments that would be allowed by the clause, yeah?

it seems to only add MINGW to the list of environments that would be allowed by the clause

Right.

The logic of change to line 533 leaves possibility of Unix, Win64, Linux, et al.

I think that's good though, how does that leave out Unix and stuff, since the (NOT WIN32 AND NOT APPLE) clause is still true for those?

@Biswa96, @longnguyen2004, would this be compatible/consistent with #3068?

Yes it's fine, this is for install directories, and the other one is for better dependency management.

Thanks, @longnguyen2004, and sorry for the long delay.