c++/10888: [3.3/3.4 regression] [diagnostic] inlining failure for allocate in ~vector()

[Gabriel Dos Reis]

Wolfgang Bangerth <bangerth@ices.utexas.edu> writes:

| > Actually I thought -Winline did not work always in 3.2 so is this a
| > "real" regression?
| 
| It's at least a regression inasmuch 
| - the user can't do anything about functions in libstdc++ being inlined or 
|   not
| - can't work around this in user code
| - will get annoying warnings
| - will probably switch off -Winline
| - and then miss things in his own code.

I can't qualify this a regression since -Winline never really worked
in previous releases (and that was broken for ages).  And this warning
really depends on the strategy used for inlining, bottom-up or
top-down or mixed.  Maybe we should not be warning if in system headers
(that is just silly).  But apart from that, I don't think we could do
anything until we settle the inlining issue.  I would suggest

  gdr -> unassigned

thanks for your analysis.

-- Gaby