Re: debug mode performance patch

[François Dumont <francois dot cppdevs at free dot fr>]

Hi

Here is the cleaned version with moreover an updated ChanLog.entry to 
signal some fixes in debug forward_list implementation. I simply fix the 
uglification but haven't use the skip list for the moment, I am not sure 
it will match my needs. O(log(n)) is not too bad but if you have 
thousands of vector instances in your application it will start to make 
research not negligible.

Moreover I forget to signal that in _Safe_sequence_base::_M_swap I had 
to lock both sequences mutex. So I hope that the scope_lock are 
reentrant cause for the moment both sequences might be associated to the 
same mutex instance. I hope I haven't been simply lucky during tests.

Regards


On 11/02/2010 09:55 PM, François Dumont wrote:
> On 11/02/2010 11:17 AM, Paolo Carlini wrote:
> > This morning I wondered if we could learn something from this work,
> > which we have been able to apply without breaking the binary 
> > compatibility:
> >
> >      http://gcc.gnu.org/ml/libstdc++/2007-06/msg00017.html
> >
> > (see also the ensuing exchanges until the final commit in October,
> > 2007). Maybe Jakub is willing to have a look too, I'm adding him in CC.
> >
> > As a minor nit, I see some unuglified names, like transfered_iterators.
>
> Ok, I will take care of it and resubmit the patch before the end of 
> the week.
>
> François

Attachment: ChangeLog.entry
Description: Text document

Attachment: performance.patch
Description: Text document