Message ID | 20211005083512.GT304296@tucnak |
---|---|
State | New |
Headers | show |
Series | c++: Implement C++23 P2334R1 - #elifdef/#elifndef | expand |
On Tue, Oct 05, 2021 at 10:35:12AM +0200, Jakub Jelinek wrote: > Hi! > > This patch implements C++23 P2334R1, which is easy because Joseph has done > all the hard work for C2X already. > Unlike the C N2645 paper, the C++ P2334R1 contains one important addition > (but not in the normative text): > "While this is a new preprocessor feature and cannot be treated as a defect > report, implementations that support older versions of the standard are > encouraged to implement this feature in the older language modes as well > as C++23." > so there are different variants how to implement it. > One is in the patch below, ignores that sentence and only implements it > for -std=c++23/-std=gnu++23 like it is only implemented for -std=c23. > Another option would be to implement it also in the older GNU modes but > not in the C/CXX modes (but it would be strange if we did that just for > C++ and not for C). > Yet another option is to enable it unconditionally. This would work for me, but... > And yet another option would be to enable it unconditionally but emit > a warning (or pedwarn) when it is seen. ...frankly, I'd prefer this last option: enable it unconditionally but emit a pedwarn when -Wpedantic. I think let's see what Joseph and Jason think before you change your patch though; don't want to add more work for you. > Note, when it is enabled for the older language modes, as Joseph wrote > in the c11-elifdef-1.c testcase, it can result e.g. in rejecting previously > valid code: > #define A > #undef B > #if 0 > #elifdef A > #error "#elifdef A applied" > #endif > #if 0 > #elifndef B > #error "#elifndef B applied" > #endif > Note, seems clang went the enable it unconditionally in all standard > versions of both C and C++, no warnings or anything whatsoever, so > essentially treated it as a DR that changed behavior of e.g. the above code. > > 2021-10-05 Jakub Jelinek <jakub@redhat.com> > > libcpp/ > * init.c (lang_defaults): Implement P2334R1, enable elifdef for > -std=c++23 and -std=gnu++23. > gcc/testsuite/ > * g++.dg/cpp/elifdef-1.C: New test. > * g++.dg/cpp/elifdef-2.C: New test. > * g++.dg/cpp/elifdef-3.C: New test. > > --- libcpp/init.c.jj 2021-09-02 10:01:15.954715595 +0200 > +++ libcpp/init.c 2021-10-05 09:55:15.010620700 +0200 > @@ -122,8 +122,8 @@ static const struct lang_flags lang_defa > /* CXX17 */ { 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 1, 0, 0, 0 }, > /* GNUCXX20 */ { 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 0, 0 }, > /* CXX20 */ { 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 0, 0 }, > - /* GNUCXX23 */ { 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 0 }, > - /* CXX23 */ { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 0 }, > + /* GNUCXX23 */ { 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1 }, > + /* CXX23 */ { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1 }, > /* ASM */ { 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } > }; > > --- gcc/testsuite/g++.dg/cpp/elifdef-1.C.jj 2021-10-05 10:00:41.410057024 +0200 > +++ gcc/testsuite/g++.dg/cpp/elifdef-1.C 2021-10-05 10:00:33.110173069 +0200 > @@ -0,0 +1,3 @@ > +// { dg-do preprocess { target { ! c++23 } } } > + > +#include "../../gcc.dg/cpp/c11-elifdef-1.c" > --- gcc/testsuite/g++.dg/cpp/elifdef-2.C.jj 2021-10-05 10:01:30.345372808 +0200 > +++ gcc/testsuite/g++.dg/cpp/elifdef-2.C 2021-10-05 10:03:36.560608083 +0200 > @@ -0,0 +1,4 @@ > +// P2334R1 > +// { dg-do preprocess { target c++23 } } > + > +#include "../../gcc.dg/cpp/c2x-elifdef-1.c" > --- gcc/testsuite/g++.dg/cpp/elifdef-3.C.jj 2021-10-05 10:01:36.029293338 +0200 > +++ gcc/testsuite/g++.dg/cpp/elifdef-3.C 2021-10-05 10:03:48.896435601 +0200 > @@ -0,0 +1,4 @@ > +// P2334R1 > +// { dg-do preprocess { target c++23 } } > + > +#include "../../gcc.dg/cpp/c2x-elifdef-1.c" > > Jakub > Marek
On Tue, 5 Oct 2021, Jakub Jelinek via Gcc-patches wrote: > One is in the patch below, ignores that sentence and only implements it > for -std=c++23/-std=gnu++23 like it is only implemented for -std=c23. > Another option would be to implement it also in the older GNU modes but > not in the C/CXX modes (but it would be strange if we did that just for > C++ and not for C). > Yet another option is to enable it unconditionally. > And yet another option would be to enable it unconditionally but emit > a warning (or pedwarn) when it is seen. > Note, when it is enabled for the older language modes, as Joseph wrote > in the c11-elifdef-1.c testcase, it can result e.g. in rejecting previously > valid code: It would probably be reasonable to enable it in older GNU modes for C as well as C++ if desired (and, in that case, emit a pedwarn-if-pedantic when it's acted on) - cases where it affects compatibility should be rare. Enabling with a pedwarn in strict modes is problematic because it changes semantics of valid code where it was inside #if 0, however. It doesn't make sense at all to me to think of a new feature like this (one with no prior art in C mentioned in the WG14 proposal) as a defect fix. Any normal directive - i.e. one that has no effect on the preprocessor #if structure and so is ignored inside #if 0 for all language versions - can more reasonably be enabled for all language versions with a pedwarn when used for old versions. (In particular, that will be appropriate for #warning, where the "don't pedwarn in C2X modes" part needs implementing after N2686 was accepted at the August / September WG14 meeting - I don't know if C++ is doing anything with #warning.)
--- libcpp/init.c.jj 2021-09-02 10:01:15.954715595 +0200 +++ libcpp/init.c 2021-10-05 09:55:15.010620700 +0200 @@ -122,8 +122,8 @@ static const struct lang_flags lang_defa /* CXX17 */ { 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 1, 0, 0, 0 }, /* GNUCXX20 */ { 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 0, 0 }, /* CXX20 */ { 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 0, 0 }, - /* GNUCXX23 */ { 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 0 }, - /* CXX23 */ { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 0 }, + /* GNUCXX23 */ { 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1 }, + /* CXX23 */ { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1 }, /* ASM */ { 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }; --- gcc/testsuite/g++.dg/cpp/elifdef-1.C.jj 2021-10-05 10:00:41.410057024 +0200 +++ gcc/testsuite/g++.dg/cpp/elifdef-1.C 2021-10-05 10:00:33.110173069 +0200 @@ -0,0 +1,3 @@ +// { dg-do preprocess { target { ! c++23 } } } + +#include "../../gcc.dg/cpp/c11-elifdef-1.c" --- gcc/testsuite/g++.dg/cpp/elifdef-2.C.jj 2021-10-05 10:01:30.345372808 +0200 +++ gcc/testsuite/g++.dg/cpp/elifdef-2.C 2021-10-05 10:03:36.560608083 +0200 @@ -0,0 +1,4 @@ +// P2334R1 +// { dg-do preprocess { target c++23 } } + +#include "../../gcc.dg/cpp/c2x-elifdef-1.c" --- gcc/testsuite/g++.dg/cpp/elifdef-3.C.jj 2021-10-05 10:01:36.029293338 +0200 +++ gcc/testsuite/g++.dg/cpp/elifdef-3.C 2021-10-05 10:03:48.896435601 +0200 @@ -0,0 +1,4 @@ +// P2334R1 +// { dg-do preprocess { target c++23 } } + +#include "../../gcc.dg/cpp/c2x-elifdef-1.c"