Re: [3.0 regression] libstdc++ and strstr

[Loren James Rittle <rittle at latour dot rsch dot comm dot mot dot com>]

>| Here is the technique: add 'extern "C"' over the definition that
>| matches the standard C header version.  The C++ standard says that
>| names defined by C may have extern "C" linkage even when provided by
>| the new style of header names and in namespace std.

> Actually it says that it is -unspecified- whether those names have C
> linkage.

Perhaps my use of English is different than yours.  But the words:
``may have'' implies ``it is possible to be this way'' not ``must be
this way''.  I.e. that choice of words is not incompatible with
``unspecified''.  For you C++ language lawyers out there (and Gabriel
is one of the best of them ;-), I am quite sorry I didn't quote
chapter and verse from ISO-C++-14882 section 17.4.2.2 clause 2 which I
had consulted just prior to posting my analysis of the bug reported by
Gerald:

``It is unspecified whether a name from the Standard C library
  declared with external linkage has either extern "C" or extern "C++"
  linkage.162) (footnote 162 reads: The only reliable way to declare
  an object or function signature from the Standard C library is by
  including the header that declares it, notwithstanding the latitude
  granted in subclause 7.1.7 of the C Standard.)"

> I don't think it allows users to redeclare those names.

The C++ library implementor is not a normal user.  To be completely
clear: I was referring to a change that we would be allowed to make as
libstdc++-v3 implementors.  I am sorry if this wasn't clear from my
initial post.  I assume that people discussing Gerald's bug report on
gcc-bugs will want to attempt to figure out how we (implementors not
end-users) can change libstdc++-v3 before 3.0 is released to fix this
regression from 2.95's libstdc++.

My analysis is that there is no easy quick-fix possible without
violating at least some aspect of the C++ standard.  If you have some
analysis of this issue beyond what I posted or another angle to get to
a solution before finishing the shadow header work, then *please* post
it instead of nit-picking my analysis. ;-)

Regards,
Loren