Re: [boost] [Stacktrace] Second review begins today 17th Mar ends 26th Mar

Subject: Re: [boost] [Stacktrace] Second review begins today 17th Mar ends 26th Mar
From: Peter Dimov (lists_at_[hidden])
Date: 2017-03-20 20:52:31

Antony Polukhin wrote:

> In the C++ standard there is no requirement that pointer to frame is a 
> pointer to function. C++ Standard does not mention frames at all and many 
> OS return frames as just a void*, not void(*)().

That's true but you've stated that you don't want to support manually 
captured frames. So what the OS returns should be irrelevant in principle - 
it's an implementation detail that the library hides from the user.

You do support frames constructed from function pointers, such as the one 
returned from ::signal, so this is what the constructor should take.

The only reason to take void* would be to support passing the return value 
of dlsym to the constructor without a reinterpret_cast.

> I'll add a `template <class T> frame(T* addr)`

This will do if you also constrain it to is_function<T>.