This is the mail archive of the
mailing list for the GCC project.
Re: Re: Errors building mozilla thunderbird 38.7.0 with gcc 6
- From: "lh_mouse"<lh_mouse at 126 dot com>
- To: "Jonathan Wakely"<jwakely dot gcc at gmail dot com>, "Daniel Letai"<dani at letai dot org dot il>
- Cc: "gcc-help"<gcc-help at gcc dot gnu dot org>
- Date: Thu, 5 May 2016 17:53:33 +0800
- Subject: Re: Re: Errors building mozilla thunderbird 38.7.0 with gcc 6
- Authentication-results: sourceware.org; auth=none
- References: <572B0C75 dot 7050001 at letai dot org dot il> <CAH6eHdT36-JciOE-arq0fDA=QcivZ9por7FpVK5eNkE=cUtZPg at mail dot gmail dot com><CAH6eHdTWWu4SXJUNvKivdx2bPtFCw1p75hTPO2JbJky6x4gSuQ at mail dot gmail dot com>
FWIW, I think it is problematic to #include_next<stdlib.h> in <cstdlib>.
It isn't <cstdlib> that should #include<stdlib.h> (it should be using a plain #include<stdlib.h>). It is <stdlib.h> that should #include_next<stdlib.h> because they have the same name.
The implementation at the moment is proved to cause a relocation problem. There has been nobody willing to solve it yet.
It is still unconfirmed whether it is a c pre-processor bug or a driver bug, but, AFAIC, if it turns out to be neither, failure to find "stdlib.h" from the C library of the #include_next directive in <cstdlib> will definitely be a libstdc++ bug.
åääïJonathan Wakely <firstname.lastname@example.org>
äéïRe: Errors building mozilla thunderbird 38.7.0 with gcc 6
On 5 May 2016 at 10:08, Jonathan Wakely wrote:
> On 5 May 2016 at 10:03, Daniel Letai wrote:
>> I could be wrong but it appears there are 2 distinct errors - in gcc6 itself
>> (specifically in header files) and in thunderbird (although this version
>> does compile with gcc 4.8.2)
>> Background - rebuilding the thunderbird 38.7.0 rpm (from redhat) using gcc6
>> and binutild 2.26 in RHEL 6.7
>> I'm wandering if this is a result of moving to std-14 as default and thus
>> not a bug, or if there is some issue with stdlib.h/cstdlib ?
> The heaer problem is a bug in the mozilla code, see
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70722 and
And the other problem is probably due to not finding the right
headers, so is the same problem.
If not, then the call to abs(sl) might need to be adjusted to pass an
integer, not a double.
Both of these issues are documented at https://gcc.gnu.org/gcc-6/porting_to.html