r23647 breaks boinc_api.h for C

[MattArsenault]

r23647 introduced this into the BOINC API header. Since bool doesn't exist in C, this doesn't work. I don't know what the purpose of these are, and can't actually find them used in the source. They could move somewhere else (perhaps down to the implementation stuff), but another alternative is to just use int. Tangentially related, the graphics API should use C linkage.

#ifdef _WIN32

Jason: New windows exit handling to play more nicely with cuda & MS-CRT extern volatile bool worker_thread_exit_request; extern volatile bool worker_thread_exit_ack;

#endif

[MattArsenault]

Replace bool with int

[MattArsenault]

Use extern C for graphics API

[davea]

The first problem was already fixed.

Is there any reason to use extern C for the graphics API? i.e. is there any project that writes its graphics app in C (and if so, why?)

[MattArsenault]

I just checked the current r23698 and it's still there.

A more appropriate question would be why would it have C++ linkage. Those functions don't actually need any C++ features, so the only thing these are getting are mangled names, needlessly breaking linking from C as well as linking between binaries from different compilers (i.e. on Windows mixing different libraries compiled with MinGW and Visual Studio). It not only breaks usage from C, but if something were to be written in any other language, C FFIs are much more common and work much better. C++ is a bad language that adds a massive amount of complexity with minimal benefit, so plenty of people stick with C.

[davea]

Oops, I forgot to check in boinc_api.h

Anti-C++ rants aside, has any project written or wanted to write a graphics module in C?

[MattArsenault]

I've been working one one. Right now I have a wrapper calling the actual functions, and it's just kind of dumb.