On 05.06.2013 08:04, Walter Bright wrote:
> On 6/4/2013 10:25 PM, dennis luehring wrote:
>> Am 05.06.2013 07:14, schrieb dennis luehring:
>>> Am 04.06.2013 19:52, schrieb Walter Bright:
>>>> On 6/4/2013 2:16 AM, dennis luehring wrote:
>>>>> in dmd/src/vcbuild/warnings.h
>>>>>
>>>>> is it realy a good idea to block the warning from popping up then
>>>>> to fix the
>>>>> stuff or are all warnings non fixables?
>>>>
>>>>
>>>> Not all VC warnings are good ideas.
>>>>
>>>
>>> examples?
>>>
>>
>> better asked: why global switch of the warnings, are all occurences
>> just wrong?
>>
>>
>
> The file was written by Rainer Schuetze and Mark Mohr. I don't know the
> specifics, except that none of them in warnings.h found real errors, and
> frankly were far more annoying than useful when I tried VC++ on dmd with
> all warnings on.

Most of the warnings relate to coding style (like warning about unreferenced local variables, signed/unsigned conflicts, etc which Walter often expressed not wanting in D, so I assumed that's even more in C). Others are caused by massive usage of the C preprocessor (like casting constants or using them like normal variables causing the flow analysis to produce warnings).

I'm undecided about using bools in arithmetic expression, disabling these warnings was added by Mark.

Leaving warnings on, but having no good reason to fix them (but to mute the compiler) just leads to ignore every warning, and that seems more dangerous.