LWG 2260 and 'r'

[timsong-cpp]

The P/R of LWG2260 (adopted in Kona '17) was written when the descriptive variable definitions (Table 27 in N4140) included a variable called r.

Then d2afc9a editorially removed that variable as unused, but the P/R of the issue was never updated. So now the allocator requirements table contains a dangling reference.

[jwakely]

Adding a row for r pushes the last row onto a new page, and then there's a \newpage, which creates an almost empty page. Removing the page break seems to be OK. @tkoeppe was that the one added because of the weird table spilling into the footer?

An alternative might be to use *p instead of r. That would be an expression of the correct type, but looks a bit strange to state the requirement on pointer_to in terms of a X::pointer object, because you wouldn't need to use pointer_to if you had p already.

[tkoeppe]

Feel free to put the r back in and remove the \newpage. The latter is situational.

[jwakely]

I'm leaning towards using *p instead of r, and adding an assertion that the result is equal to p.