Re: Bug with GCC's handling of lifetimes of implicit-lifetime types

[Jonathan Wakely <jwakely.gcc@gmail.com>]

On Sun, 11 Dec 2022 at 13:44, Gavin Ray wrote:
>
> Whoops, the last line should be pages_span(storage, ...)

That won't even compile, so there's your main difference.

That seems even more off-topic on this list than "somebody told me gcc
is broken, because of [unrelated code example with UB]".

A creates 'pages' objects in the storage, B doesn't compile, but even
if it did, it's just saying "look at this storage as though it
contained this many objects of type 'pages'" which doesn't create any
objects, and would be a lie.


>
> On Sun, Dec 11, 2022 at 8:38 AM Gavin Ray <ray.gavin97@gmail.com> wrote:
>>
>> @Richard
>>
>> That's some intense code, I appreciate the code-samples and explanation, thank you =)
>>
>> @Jonathan
>>
>> Maybe there was some misunderstanding?
>> I didn't make the connection either but I also don't know that much about C++
>>
>> It seems like that expression is valid then? Good to know =)
>>
>> As a random aside if I may -- what is the difference between placement-new of pointers in
>> std::byte storage, and making a std::span over the storage area?
>>
>> std::byte storage[PAGE_SIZE * NUM_PAGES];
>>
>> // A)
>> page* pages = new (storage) page[NUM_PAGES];
>> // B)
>> std::span<page, NUM_PAGES> pages_span(pages, NUM_PAGES);