[Lldb-commits] [PATCH] Implements a HostThread class.

[Zachary Turner]

That's actually exactly what I've done!   Check the code in HostThread.h in
the original patch.  The difference is that instead of empty inheritance, I
used a typedef.  But the idea is exactly the same.


On Thu, Aug 28, 2014 at 8:53 PM, Bruce Mitchener <bruce.mitchener at gmail.com>
wrote:

> One approach that I've seen for this in games where virtual function calls
> are an unhappy thing is a pattern where the inheritance appears to be
> inverted:
>
> In this case, you'd have each platform class in their own files and
> directory structure:
>
>     class HostThreadMacOSX { ... };
>
>     class HostThreadWindows { ... };
>
> Then, you'd have a single HostThread.h that looked roughly like:
>
>     if (__WIN32__)
>     #include "threading/windows/HostThreadWindows.h"
>
>     namespace lldb
>     {
>     class HostThread : public Windows::HostThreadWindows
>     {
>     };
>     }
>     #elif __APPLE__
>     #include "threading/macosx/HostThreadMacOSX.h"
>     namespace lldb
>     {
>     class HostThread : public MacOSX::HostThreadMacOSX
>     {
>     };
>     }
>     #else
>     #error "lldb::HostThread not implemented on this platform!"
>     #endif
>
> Now, all of the code can reference HostThread, there are no virtual calls,
> and each platform can invoke platform-specific behavior with ease.
>
> It has some downsides, like the exact interface of what is required isn't
> as clear (since there's no "interface" class), but I've had good luck with
> this structure in the past.
>
> Hope this is interesting / useful!
>
>  - Bruce
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/lldb-commits/attachments/20140828/b8d4c096/attachment.html>