News for hackers. Stuff that matters.
Re:Some Were Meant For C (2017) [pdf] by qznc on Monday March 01, 2021 @02:55PM (#26307519)
You would constraint yourself to a subset of C++ which is not C. For example, forbid all raw pointers and only use smart pointers. Now your memory is freed automatically via RAII and use-after-free errors are much harder to create.
Re:Some Were Meant For C (2017) [pdf] by coliveira on Monday March 01, 2021 @03:11PM (#26307740)
As you said, this is a subset of C++. But everyone debates what the right subset should be. In reality C++ is incapable of solving the problem, it can only propose good practices.
Re:Some Were Meant For C (2017) [pdf] by aromatic_dev on Monday March 01, 2021 @05:33PM (#26309396)
Though if one only uses smart pointers they will have to accept the performance hit that that brings -- in high performance situations, this may be unacceptable.
Re:Some Were Meant For C (2017) [pdf] by ncmncm on Monday March 01, 2021 @06:43PM (#26310070)
Circumstances where using smart pointers affects performance are vanishingly rare. Where performance matters, on any critical path no allocation or deallocation is done, so a smart pointer can have no effect. On other paths, any difference is swamped by overhead in malloc and free.

So, in actual practice, no.

Re:Some Were Meant For C (2017) [pdf] by wheybags on Monday March 01, 2021 @06:58PM (#26310196)
There's some argument to be made about bookkeeping and (maybe? I'm not sure) an extra level of indirection with shared_ptr, but the cost of unique_ptr is literally zero.
Re:Some Were Meant For C (2017) [pdf] by ncmncm on Monday March 01, 2021 @08:43PM (#26310925)
Passing unique_ptr by move works out to be surprisingly expensive, so one doesn't, where that matters. (It is usually a mistake to have unique_ptr visible in an interface at all, but passing a pimpl type, implemented with unique_ptr, by move costs exactly the same.)

And of course no one who cares for their reputation is seen using shared_ptr.

Re:Some Were Meant For C (2017) [pdf] by boring_twenties on Tuesday March 02, 2021 @12:22AM (#26312155)
What's the expense of moving unique_ptr -- is it resetting the source to nullptr?
Re:Some Were Meant For C (2017) [pdf] by mawww on Tuesday March 02, 2021 @01:35AM (#26312480)
My guess is that because in most common x86_64 calling conventions the caller is responsible for destructing the parameters, it has to run the unique_ptr parameter destructor which needs to check if it was moved from in the callee or not. Additionally, because this destructor is not a trivial one (i.e. it does something), unique_ptr cannot be passed directly through a register but must be spilled on the stack.
Re:Some Were Meant For C (2017) [pdf] by ncmncm on Tuesday March 02, 2021 @09:40AM (#26315754)
This is correct. Chandler Carruth gave a talk about unfortunate pessimizations imposed by the ABI. Where the call is inlined, that stuff is optimized out, but functions that take a moved unique_ptr or pimpl object -- taking ownership of a heap object -- tend not to be inlined unless they mostly just pass it along to one that isn't inlined.

So, the cost of moving a smart pointer onto the stack and, presumably, off again to someplace less ephemeral might matter, on a critical path. But it would be a mistake to exaggerate this: we are talking about a couple of unfortunate ordinary memory operations, not traps, function calls, or atomic synchronizations, when you might have hoped to be using just registers. If you are already touching memory -- something pretty common -- it is just more of that.

Re:Some Were Meant For C (2017) [pdf] by wheybags on Tuesday March 02, 2021 @10:03AM (#26315995)
Wow, did not know this, I'm going to go and have a deeper look at it now. Thanks to all in this comment thread!