This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH][C++] Fix PR29433, make C++ use a lot less time/memory


On Wed, Feb 21, 2007 at 04:45:47PM +0100, Michael Matz wrote:
> Anyway, now the differences between current trunk and patched trunk look 
> like so:
> 
> -       .string "basic_streambuf<char,std::char_traits<char> >"
> +       .string "basic_streambuf<char, std::char_traits<char> >"
> -       .string "list<QString,std::allocator<QString> >"
> +       .string "list<QString, std::allocator<QString> >"
> 
> So, it's just spacing.  

For the record, this sort of change can still cause problems - but (A)
it looks like the new output is less problematic than the old output,
and (B) we're so far into GDB Sucks At Templates territory at this
point that I just don't care.  There are currently problems wherever
the names don't match what the demangler would do, because of stupid
string comparisons.  But I have had the code ready to deal with that
for years - just haven't found a way to hook it up without a big
performance hit.

So, I'm happy.

> I've also checked the gdb testsuite again with that patch, and there are 
> 14 failures less.  Compared to gdb testsuite with unpatched trunk there's 
> only one regression where it expected something different for a tagless 
> struct:

I'm happy for us to fix that one in the GDB testsuite; the new output
is more sensible anyway.  If the patch is otherwise OK, I have no
objections any more.  Thank you for going the extra mile on this!

-- 
Daniel Jacobowitz
CodeSourcery


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]