<p dir="ltr"><br>
El 31/7/2015 2:37, "Sebastian Huber" <<a href="mailto:sebastian.huber@embedded-brains.de">sebastian.huber@embedded-brains.de</a>> escribió:<br>
><br>
> Hello Daniel,</p>
<p dir="ltr">Hello Sebastian.</p>
<p dir="ltr">><br>
><br>
> On 30/07/15 17:26, Daniel Gutson wrote:<br>
>><br>
>> On Thu, Jul 30, 2015 at 11:31 AM, Daniel Gutson<br>
>> <<a href="mailto:daniel.gutson@tallertechnologies.com">daniel.gutson@tallertechnologies.com</a>> wrote:<br>
>>><br>
>>> ><br>
>>> >El 30/7/2015 11:27, "Joel Sherrill"<<a href="mailto:joel.sherrill@oarcorp.com">joel.sherrill@oarcorp.com</a>> escribió:<br>
>>>><br>
>>>> >><br>
>>>> >><br>
>>>> >><br>
>>>> >>On 7/30/2015 9:08 AM, Daniel Gutson wrote:<br>
>>>>><br>
>>>>> >>><br>
>>>>> >>>IOW, I think that the double parens is only for decltype.<br>
>>>><br>
>>>> >><br>
>>>> >><br>
>>>> >>Historical convention is to put parens around variable names<br>
>>>> >>in macros. What type of impact does this have?<br>
>>><br>
>>> ><br>
>>> >If what I think is correct, then the impact os none since this a bug. But I<br>
>>> >will look at it deeper once I arrive to the office in 1h. I will try with<br>
>>> >different versions of gcc and clang and look into the C++ standard. So far<br>
>>> >the (()) seems to be a decltype only thing, so this would be a frontend bug.<br>
>><br>
>> As I mentioned in the bugzilla, I think this is a bug of the front-end<br>
>> (since I could not reproduce<br>
>> it in earlier versions of g++ (4.8.4) and clang (3.5)). I already<br>
>> asked Ville Voutilainen and Jens Maurer<br>
>> (from the C++ Committee) to look into it. I will let you know.<br>
><br>
><br>
> thanks for looking at this issue. Is this really a problem within the scope of the standard? </p>
<p dir="ltr">The bug has been introduced when implementing the new rules related to decltype in C++14. Indeed the code has something like 'if (cxxversion < CPP14) return;' meaning that the newly (faulty) introduced code is not executed when running previous std versions: that's why -std=c++11 should work (the problem is that g++ 6 defaults to c++14).</p>
<p dir="ltr">>The global register variables are GCC specific.</p>
<p dir="ltr">Yes, a GNU C extension.</p>
<p dir="ltr">> Is it possible to construct objects without an address via plain C++?</p>
<p dir="ltr">Sorry I don't understand the question. Rephrase please?</p>
<p dir="ltr">Global objects and objects of static storage duration don't take an address.<br></p>
<p dir="ltr">><br>
><br>
> -- <br>
> Sebastian Huber, embedded brains GmbH<br>
><br>
> Address : Dornierstr. 4, D-82178 Puchheim, Germany<br>
> Phone : +49 89 189 47 41-16<br>
> Fax : +49 89 189 47 41-09<br>
> E-Mail : <a href="mailto:sebastian.huber@embedded-brains.de">sebastian.huber@embedded-brains.de</a><br>
> PGP : Public key available on request.<br>
><br>
> Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.<br>
><br>
</p>