Re: [RFC] Dealing with ODR violations in GCC

2015-03-24 Thread Martin Liška
On 02/13/2015 07:54 PM, Martin Liška wrote: On 01/15/2015 06:47 AM, Jan Hubicka wrote: Hello. I've just finished successfully chromium LTO build and there's list of mainly -Wodr warnings. I think some of them are false positives? What of those you consider to be false? I wonder if we can pri

Re: [RFC] Dealing with ODR violations in GCC

2015-02-16 Thread Jan Hubicka
> On 02/16/2015 07:27 PM, Jan Hubicka wrote: > >Jason, I wonder if there is more informative way to print destructor during > >LTO than as __comp_dtor? > > You could print the mangled name, and possibly run it through > __cxa_demangle if it starts with _Z. > > >Laos for named types, > >perhaps p

Re: [RFC] Dealing with ODR violations in GCC

2015-02-16 Thread Jason Merrill
On 02/16/2015 07:27 PM, Jan Hubicka wrote: Jason, I wonder if there is more informative way to print destructor during LTO than as __comp_dtor? You could print the mangled name, and possibly run it through __cxa_demangle if it starts with _Z. Laos for named types, perhaps printing just typ

Re: [RFC] Dealing with ODR violations in GCC

2015-02-16 Thread Jan Hubicka
Hi, the warning about types of fields seems misplaced: > ../../third_party/WebKit/public/platform/WebCryptoAlgorithmParams.h:295:0: > warning: type ???struct WebCryptoEcKeyImportParams??? violates one definition > rule [-Wodr] > class WebCryptoEcKeyImportParams : public WebCryptoAlgorithmParams

Re: [RFC] Dealing with ODR violations in GCC

2015-02-13 Thread Martin Liška
On 01/15/2015 06:47 AM, Jan Hubicka wrote: >> Hello. >> >> I've just finished successfully chromium LTO build and there's list of >> mainly -Wodr warnings. >> I think some of them are false positives? > > What of those you consider to be false? > I wonder if we can print just type name so we avoi

Re: [RFC] Dealing with ODR violations in GCC

2015-01-14 Thread Jan Hubicka
> Hello. > > I've just finished successfully chromium LTO build and there's list of mainly > -Wodr warnings. > I think some of them are false positives? What of those you consider to be false? I wonder if we can print just type name so we avoid using the wrong "struct" in place of class... > >

Re: [RFC] Dealing with ODR violations in GCC

2015-01-14 Thread Martin Liška
On 09/12/2014 07:40 AM, Jan Hubicka wrote: Hi, I went through excercise of running LTO bootstrap with ODR verification on. There are some typename clashes I guess we want to fix. I wonder what approach is preferred, do we want to introduce anonymous namespaces for those? Honza ../../gcc/tli

Re: [RFC] Dealing with ODR violations in GCC

2014-09-24 Thread Jonathan Wakely
On 12 September 2014 06:40, Jan Hubicka wrote: > Hi, > I went through excercise of running LTO bootstrap with ODR verification on. > There are some typename clashes > I guess we want to fix. I wonder what approach is preferred, do we want to > introduce anonymous > namespaces for those? > /usr/

Re: [RFC] Dealing with ODR violations in GCC

2014-09-12 Thread Trevor Saunders
On Fri, Sep 12, 2014 at 10:16:12AM +0200, Richard Biener wrote: > On Fri, Sep 12, 2014 at 7:40 AM, Jan Hubicka wrote: > > Hi, > > I went through excercise of running LTO bootstrap with ODR verification on. > > There are some typename clashes > > I guess we want to fix. I wonder what approach is

Re: [RFC] Dealing with ODR violations in GCC

2014-09-12 Thread Richard Biener
On Fri, Sep 12, 2014 at 7:40 AM, Jan Hubicka wrote: > Hi, > I went through excercise of running LTO bootstrap with ODR verification on. > There are some typename clashes > I guess we want to fix. I wonder what approach is preferred, do we want to > introduce anonymous > namespaces for those?