Re: Semicolons at the end of member function definitions

2007-08-03 Thread Gabriel Dos Reis 
On Wed, 1 Aug 2007, Mark Mitchell wrote:

| Volker Reichelt wrote:
| 
|  2007-03-26  Dirk Mueller  [EMAIL PROTECTED]
|  
| * parser.c (cp_parser_member_declaration): Pedwarn
| about stray semicolons after member declarations.
|  
| 
|  It makes
|  
|struct A
|{
|   void foo() {};
|}
| 
| That is indeed still legal in the current working draft.  (The reason
| that I copied the grammar productions above the parser functions was so
| that it would be easy to check things like this...)
| 
|  Therefore, IMHO the patch is wrong and should be reverted.
| 
| Yes, please go ahead and revert it.  And, if you have time, please add a
| test-case specifically for this case.  The previous patch removed
| semicolons from lots of valid code, but probably none of those test
| cases were specifically for this case.

I agree with Mark.  
Sorry for the confusion.

-- Gaby

Re: Semicolons at the end of member function definitions

2007-08-02 Thread Dirk Mueller 
On Tuesday, 31. July 2007, Volker Reichelt wrote:

 Therefore, IMHO the patch is wrong and should be reverted.
 Or am I missing something?

I don't have access to a newer version of the language standard either. I 
couldn't find a different case that is comparable to this one either (for 
example, we handle the difference between an enumerator list with a 
trailing ',' and a initializer list with a trailing ',' - although that is 
equally counter-intuitive). 

So, assuming that this hasn't changed in a newer standard then I guess you're 
right and the check should only be a warning with -pedantic, (and 
no -pedantic-errors) given. This is however not implementable currently. As I 
don't know why -pedantic-errors is on by default for C++ (but not for C), the 
only possibility I see is indeed to revert this patch and add a 
comment/testcase. 


Greetings,
Dirk

Re: Semicolons at the end of member function definitions

2007-08-02 Thread Manuel López-Ibáñez 
On 01/08/07, Dirk Mueller [EMAIL PROTECTED] wrote:
 On Tuesday, 31. July 2007, Volker Reichelt wrote:

 So, assuming that this hasn't changed in a newer standard then I guess you're
 right and the check should only be a warning with -pedantic, (and
 no -pedantic-errors) given. This is however not implementable currently. As I
 don't know why -pedantic-errors is on by default for C++ (but not for C), the
 only possibility I see is indeed to revert this patch and add a
 comment/testcase.

pedantic-errors is not on by default in C++. The function pedwarn()
produces errors by default in C++ as opposed to C. On the other hand,
-pedantic-errors makes pedwarn() produce errors instead of warnings
AND sets the flag pedantic. It is a subtle difference. (Then,
-pedantic sets the flag pedantic but does not change the way
pedwarn() works and -fpermissive just change pedwarn() to emit
warnings instead of errors). This is how it works internally, now
whether it is reflecting the desired behaviour I just don't know.

With the above cheat sheet, you can play to guess what is the actual
effect of combinations like 1) -pedantic, 2) -pedantic
-fpermissive 3) -fpermissive -Werror 4) -fpermissive 5)
-pedantic-errors -fpermissive.

(As you will see, 4 and 5 are not the same, so no, -pedantic-errors is
not enabled by default in C++ however, 2 and 5 will produce the same
output in C++).

I hope this clarifies.

Cheers,

Manuel.

Re: Semicolons at the end of member function definitions

2007-08-01 Thread Mark Mitchell 
Volker Reichelt wrote:

 2007-03-26  Dirk Mueller  [EMAIL PROTECTED]
 
* parser.c (cp_parser_member_declaration): Pedwarn
about stray semicolons after member declarations.
 

 It makes
 
   struct A
   {
  void foo() {};
   }

That is indeed still legal in the current working draft.  (The reason
that I copied the grammar productions above the parser functions was so
that it would be easy to check things like this...)

 Therefore, IMHO the patch is wrong and should be reverted.

Yes, please go ahead and revert it.  And, if you have time, please add a
test-case specifically for this case.  The previous patch removed
semicolons from lots of valid code, but probably none of those test
cases were specifically for this case.

Thanks,

-- 
Mark Mitchell
CodeSourcery
[EMAIL PROTECTED]
(650) 331-3385 x713

Semicolons at the end of member function definitions

2007-07-30 Thread Volker Reichelt 
Hi,

I just stumbled over the patch

2007-03-26  Dirk Mueller  [EMAIL PROTECTED]

   * parser.c (cp_parser_member_declaration): Pedwarn
   about stray semicolons after member declarations.

which was approved by Gaby here:
http://gcc.gnu.org/ml/gcc-patches/2007-03/msg01456.html
and made it into the trunk here:
http://gcc.gnu.org/ml/gcc-cvs/2007-03/msg00841.html

It makes

  struct A
  {
 void foo() {};
  }

a hard error with -pedantic.

The 1998 version of the standard (sorry, I don't have the 2003 version
available) contains in [class.mem]:

  member-declaration:
...
function-definition ;opt
...

Therefore, IMHO the patch is wrong and should be reverted.
Or am I missing something?

Regards,
Volker