[PATCH] c++: parse error with >= in template argument list [PR105436]

[PATCH] c++: parse error with >= in template argument list [PR105436]

[Marek Polacek]

This patch fixes an oversight whereby we treated >= as the end of
a template argument.  This causes problems in C++14, because in
cp_parser_template_argument we go different ways for C++14 and C++17:

  /* It must be a non-type argument.  In C++17 any constant-expression is
     allowed.  */
  if (cxx_dialect > cxx14)
    goto general_expr;

so in this testcase in C++14 we get "N" as the template argument but in
C++17 it is the whole "N >= 5" expression.  So in C++14 the remaining
">= 5" triggered the newly-added diagnostic.

Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk?

	PR c++/105436

gcc/cp/ChangeLog:

	* parser.cc (cp_parser_next_token_ends_template_argument_p): Don't
	return true for CPP_GREATER_EQ.

gcc/testsuite/ChangeLog:

	* g++.dg/parse/template31.C: New test.
---
 gcc/cp/parser.cc                        | 1 -
 gcc/testsuite/g++.dg/parse/template31.C | 4 ++++
 2 files changed, 4 insertions(+), 1 deletion(-)
 create mode 100644 gcc/testsuite/g++.dg/parse/template31.C

diff --git a/gcc/cp/parser.cc b/gcc/cp/parser.cc
index a5cbb3e896f..5fa743b5a8e 100644
--- a/gcc/cp/parser.cc
+++ b/gcc/cp/parser.cc
@@ -33224,7 +33224,6 @@ cp_parser_next_token_ends_template_argument_p (cp_parser *parser)
 	  || ((cxx_dialect != cxx98) && token->type == CPP_RSHIFT)
 	  /* For better diagnostics, treat >>= like that too, that
 	     shouldn't appear non-nested in template arguments.  */
-	  || token->type == CPP_GREATER_EQ
 	  || token->type == CPP_RSHIFT_EQ);
 }
 
diff --git a/gcc/testsuite/g++.dg/parse/template31.C b/gcc/testsuite/g++.dg/parse/template31.C
new file mode 100644
index 00000000000..a5693e851f7
--- /dev/null
+++ b/gcc/testsuite/g++.dg/parse/template31.C
@@ -0,0 +1,4 @@
+// PR c++/105436
+
+template<bool> struct A;
+template<int N> A<N >= 5> f();

base-commit: 1cb220498e1f59021dab36c39c5d726e9f070c6a
-- 
2.35.1

Re: [PATCH] c++: parse error with >= in template argument list [PR105436]

[Jason Merrill]

On 5/2/22 12:19, Marek Polacek wrote:
> This patch fixes an oversight whereby we treated >= as the end of
> a template argument.  This causes problems in C++14, because in
> cp_parser_template_argument we go different ways for C++14 and C++17:
> 
>    /* It must be a non-type argument.  In C++17 any constant-expression is
>       allowed.  */
>    if (cxx_dialect > cxx14)
>      goto general_expr;
> 
> so in this testcase in C++14 we get "N" as the template argument but in
> C++17 it is the whole "N >= 5" expression.  So in C++14 the remaining
> ">= 5" triggered the newly-added diagnostic.

Hmm, I think >>= is questionable as well, as it could resolve to a 
constexpr operator>>=.  Seems like the two calls to 
cp_parser_next_token_ends_template_argument_p in the C++14 non-type 
argument parsing code need to be strictly correct to avoid the problem 
in this testcase; maybe we want a separate overload or another parameter 
to indicate that?

> Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk?
> 
> 	PR c++/105436
> 
> gcc/cp/ChangeLog:
> 
> 	* parser.cc (cp_parser_next_token_ends_template_argument_p): Don't
> 	return true for CPP_GREATER_EQ.
> 
> gcc/testsuite/ChangeLog:
> 
> 	* g++.dg/parse/template31.C: New test.
> ---
>   gcc/cp/parser.cc                        | 1 -
>   gcc/testsuite/g++.dg/parse/template31.C | 4 ++++
>   2 files changed, 4 insertions(+), 1 deletion(-)
>   create mode 100644 gcc/testsuite/g++.dg/parse/template31.C
> 
> diff --git a/gcc/cp/parser.cc b/gcc/cp/parser.cc
> index a5cbb3e896f..5fa743b5a8e 100644
> --- a/gcc/cp/parser.cc
> +++ b/gcc/cp/parser.cc
> @@ -33224,7 +33224,6 @@ cp_parser_next_token_ends_template_argument_p (cp_parser *parser)
>   	  || ((cxx_dialect != cxx98) && token->type == CPP_RSHIFT)
>   	  /* For better diagnostics, treat >>= like that too, that
>   	     shouldn't appear non-nested in template arguments.  */
> -	  || token->type == CPP_GREATER_EQ
>   	  || token->type == CPP_RSHIFT_EQ);
>   }
>   
> diff --git a/gcc/testsuite/g++.dg/parse/template31.C b/gcc/testsuite/g++.dg/parse/template31.C
> new file mode 100644
> index 00000000000..a5693e851f7
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/parse/template31.C
> @@ -0,0 +1,4 @@
> +// PR c++/105436
> +
> +template<bool> struct A;
> +template<int N> A<N >= 5> f();
> 
> base-commit: 1cb220498e1f59021dab36c39c5d726e9f070c6a

Re: [PATCH] c++: parse error with >= in template argument list [PR105436]

[Jakub Jelinek]

On Tue, May 03, 2022 at 04:26:51PM -0400, Jason Merrill wrote:
> On 5/2/22 12:19, Marek Polacek wrote:
> > This patch fixes an oversight whereby we treated >= as the end of
> > a template argument.  This causes problems in C++14, because in
> > cp_parser_template_argument we go different ways for C++14 and C++17:
> > 
> >    /* It must be a non-type argument.  In C++17 any constant-expression is
> >       allowed.  */
> >    if (cxx_dialect > cxx14)
> >      goto general_expr;
> > 
> > so in this testcase in C++14 we get "N" as the template argument but in
> > C++17 it is the whole "N >= 5" expression.  So in C++14 the remaining
> > ">= 5" triggered the newly-added diagnostic.
> 
> Hmm, I think >>= is questionable as well, as it could resolve to a constexpr
> operator>>=.  Seems like the two calls to

The template argument is a constant-expression and >>= can't appear non-nested
in constant-expression non-terminal, can it?
>= certainly can.

	Jakub

Re: [PATCH] c++: parse error with >= in template argument list [PR105436]

[Jason Merrill]

On 5/3/22 16:43, Jakub Jelinek wrote:
> On Tue, May 03, 2022 at 04:26:51PM -0400, Jason Merrill wrote:
>> On 5/2/22 12:19, Marek Polacek wrote:
>>> This patch fixes an oversight whereby we treated >= as the end of
>>> a template argument.  This causes problems in C++14, because in
>>> cp_parser_template_argument we go different ways for C++14 and C++17:
>>>
>>>     /* It must be a non-type argument.  In C++17 any constant-expression is
>>>        allowed.  */
>>>     if (cxx_dialect > cxx14)
>>>       goto general_expr;
>>>
>>> so in this testcase in C++14 we get "N" as the template argument but in
>>> C++17 it is the whole "N >= 5" expression.  So in C++14 the remaining
>>> ">= 5" triggered the newly-added diagnostic.
>>
>> Hmm, I think >>= is questionable as well, as it could resolve to a constexpr
>> operator>>=.  Seems like the two calls to
> 
> The template argument is a constant-expression and >>= can't appear non-nested
> in constant-expression non-terminal, can it?

Ah, true, a constant-expression is a conditional-expression, which can't 
involve an assignment operator.

Jason

Re: [PATCH] c++: parse error with >= in template argument list [PR105436]

[Marek Polacek]

On Tue, May 03, 2022 at 04:46:11PM -0400, Jason Merrill wrote:
> On 5/3/22 16:43, Jakub Jelinek wrote:
> > On Tue, May 03, 2022 at 04:26:51PM -0400, Jason Merrill wrote:
> > > On 5/2/22 12:19, Marek Polacek wrote:
> > > > This patch fixes an oversight whereby we treated >= as the end of
> > > > a template argument.  This causes problems in C++14, because in
> > > > cp_parser_template_argument we go different ways for C++14 and C++17:
> > > > 
> > > >     /* It must be a non-type argument.  In C++17 any constant-expression is
> > > >        allowed.  */
> > > >     if (cxx_dialect > cxx14)
> > > >       goto general_expr;
> > > > 
> > > > so in this testcase in C++14 we get "N" as the template argument but in
> > > > C++17 it is the whole "N >= 5" expression.  So in C++14 the remaining
> > > > ">= 5" triggered the newly-added diagnostic.
> > > 
> > > Hmm, I think >>= is questionable as well, as it could resolve to a constexpr
> > > operator>>=.  Seems like the two calls to
> > 
> > The template argument is a constant-expression and >>= can't appear non-nested
> > in constant-expression non-terminal, can it?
> 
> Ah, true, a constant-expression is a conditional-expression, which can't
> involve an assignment operator.

So do you want me to make any changes or is the patch OK as-is?

Marek

Re: [PATCH] c++: parse error with >= in template argument list [PR105436]

[Jason Merrill]

On 5/4/22 11:15, Marek Polacek wrote:
> On Tue, May 03, 2022 at 04:46:11PM -0400, Jason Merrill wrote:
>> On 5/3/22 16:43, Jakub Jelinek wrote:
>>> On Tue, May 03, 2022 at 04:26:51PM -0400, Jason Merrill wrote:
>>>> On 5/2/22 12:19, Marek Polacek wrote:
>>>>> This patch fixes an oversight whereby we treated >= as the end of
>>>>> a template argument.  This causes problems in C++14, because in
>>>>> cp_parser_template_argument we go different ways for C++14 and C++17:
>>>>>
>>>>>      /* It must be a non-type argument.  In C++17 any constant-expression is
>>>>>         allowed.  */
>>>>>      if (cxx_dialect > cxx14)
>>>>>        goto general_expr;
>>>>>
>>>>> so in this testcase in C++14 we get "N" as the template argument but in
>>>>> C++17 it is the whole "N >= 5" expression.  So in C++14 the remaining
>>>>> ">= 5" triggered the newly-added diagnostic.
>>>>
>>>> Hmm, I think >>= is questionable as well, as it could resolve to a constexpr
>>>> operator>>=.  Seems like the two calls to
>>>
>>> The template argument is a constant-expression and >>= can't appear non-nested
>>> in constant-expression non-terminal, can it?
>>
>> Ah, true, a constant-expression is a conditional-expression, which can't
>> involve an assignment operator.
> 
> So do you want me to make any changes or is the patch OK as-is?

OK.