Enable warnings for non-conforming c99/c++98 code

[geky]

This will help moving towards more strictly conforming code, which will help reduce problems with portability across compilers.

This pr is currently gcc only. If anyone knows the appropriate flags for armc5 or iar, let me know.

related: #241

[geky]

Here's a quick snippet of some new warnings:

Compile: Thread.cpp
[Warning] Thread.h@207: comma at end of enumerator list [-Wpedantic]
[Warning] MK64F12.h@383: comma at end of enumerator list [-Wpedantic]
[Warning] MK64F12.h@12280: ISO C99 doesn't support unnamed structs/unions [-Wpedantic]
[Warning] PeripheralNames.h@26: comma at end of enumerator list [-Wpedantic]
[Warning] rt_TypeDef.h@44: ISO C++ 1998 does not support 'long long' [-Wlong-long]
[Warning] rt_CMSIS.c@1149: ISO C forbids conversion of function pointer to object pointer type [-Wpedantic]
[Warning] rt_CMSIS.c@1263: ISO C forbids conversion of object pointer to function pointer type [-Wpedantic]

[geky]

Hold on, looks like this change isn't as innocent as I had hoped:

/home/travis/build/mbedmicro/mbed/hal/common/RawSerial.cpp:63:66: error: 'vsnprintf' was not declared in this scope
     int len = vsnprintf(dummy_buf, sizeof(dummy_buf), format, arg);

Looks like vsnprintf wasn't introduced into c++ until c++11. Not sure why this didn't show up in local compilation.

[0xc0170]

This pr is currently gcc only. If anyone knows the appropriate flags for armc5 or iar, let me know.

I understand this would be nice to have, but sending PR without providing the valuable vital information (how many warnings we have, are there any errors, how was this tested, all toolchains aligned, etc) does not provide much.

Hold on, looks like this change isn't as innocent as I had hoped

If its only gcc, it should be stated in the commit message . This might be considered a breaking change, should be there as well, and a reason why we are doing this. have you tested any example code ? There was problem with armcc --gnu flag that some headers from vendors require this extensions.

I would imagine this become an issue, log all warnings/errors and clean-up the base slowly. Not to change the flags, without any planning to clean the code base (how big is the log to be cleaned, who will drive this and what's the time frame).

[geky]

I tested locally with an existing program and thought this would not impact the codebase.

I was hoping to get help on aligning the tools and figured a pr would be the best way to start a conversation on adding these flags. It sounds like I made a mistake here and will move this to an issue.

[geky]

Continued over here: #2152