GCC build failed for native with your patch on 2003-09-05T11:02:39Z.

Jan Hubicka jh@suse.cz
Sun Sep 7 01:13:00 GMT 2003


> 
> On Saturday, Sep 6, 2003, at 18:01 US/Pacific, Jan Hubicka wrote:
> 
> >>>Please do,
> >>>time to go sleep for me, but I will take a look into it tommorow.
> >>>
> >>>Thank you for testing!
> >>>Honza
> >>
> >>Here it is (I am getting good at this reducing thing),
> >>I have not tested your newer patch though:
> >>
> >>template<typename _CharT>
> >>  struct __timepunct_cache
> >>  {
> >>    static const _CharT* _S_timezones[14];
> >>  };
> >>template<> const char* __timepunct_cache<char>::_S_timezones[14];
> >>template<typename _CharT>
> >>  class time_get
> >>  {
> >>  public:
> >>    typedef _CharT char_type;
> >>    void _M_extract_name(const _CharT** __names) const;
> >>    void _M_extract_via_format() const;
> >>  };
> >>
> >>template<typename _CharT>
> >>  void
> >>  time_get<_CharT>::
> >>  _M_extract_via_format() const
> >>  {
> >>      _M_extract_name(__timepunct_cache<_CharT>::_S_timezones);
> >>  }
> >>template class time_get<char>;
> >
> >Can you please send me the assembly I am supposed to get and what is
> >wrong there?  The symbol in question seems to be defined just well in  
> >my
> >version (but I have somewhat modified tree so perhaps I fixed it in
> >meantime).  I will try fresh one tomorrow.
> 
> Here is the diff of the two assembly, the good one is "+" where there  
> is a non_lazy_ptr:
Hmm, I do have non_lazy_ptr in unit-at-a-time output as well, but I do
have quite number of changes in the tree.  I will try to figure out what
fixed that tomorow, today is really bit late.
Thanks!

Honza



More information about the Gcc-regression mailing list