PR c++/11808
Gabriel Dos Reis
gdr@integrable-solutions.net
Tue Sep 2 22:02:00 GMT 2003
Mark Mitchell <mark@codesourcery.com> writes:
| On Tue, 2003-09-02 at 13:40, Gabriel Dos Reis wrote:
| > Mark Mitchell <mark@codesourcery.com> writes:
| >
| > | * cp-tree.h (KOENIG_LOOKUP_P): New macro.
| >
| > Mark,
| >
| > Thanks for addressing this issue. However, the naming KOENIG_LOOKUP_P
| > is terrible. The standard term is "Argument dependent name lookup".
|
| In late-draft versions of the standard, it was still called Koenig
| lookup. In fact, there is a section [basic.lookup.koenig] in the
| version I have. Did that change in the final version?
The "tag" for navigating in the text is [basic.lookup.koenig] but the
heading is "argument-dependent name lookup", just as we have
[basic.lookup.classref] for the heading "Class member acess" or
[conv.fpprom] for "Floating point promotion". Nowhere in the standard
text is "Koenig lookup" defined or mentioned. The history about that
lookup naming is very "surprising" :-)
| > And Andy Koenig tends to distant himself from that lookup rule.
| > You might want to converse with EDG guys (especially with Daveed) if
| > you wanted more information about the history of that lookup rule.
| > (Yes, I'm aware of the wide-spread usage "koenig lookup", but it is not
| > inappropriate for GCC source I'm afraid).
|
| If you want to change it, that's fine by me, but "Koenig" appears in
| several places in the C++ front end. It would be silly to change it one
| place, but not all of them. I don't intend to make that change.
I understand that the change should be consistent, but I was worrying
about introducing one more candidate :-)
-- Gaby
More information about the Gcc-patches
mailing list