Re: [VOTE] [RFC] 64 bit platform improvements for string length and integer

How can you reduce memory overhead if you start using 64-bit number for class names and other entities? If you have been developing this patch for months, why didn't you make it already? I'll tell you - because it's impossible. Thanks. Dmitry. On Wed, May 14, 2014 at 10:27 AM, Pierre Joye <pierre.php@gmail.com> wrote: > hi, > > On Wed, May 14, 2014 at 7:57 AM, Dmitry Stogov <dmitry@zend.com> wrote: > > Pierre, > > > > you developed a patch that will negatively affect every PHP application, > > just to use "modern" types... > > it's completely not necessary step for 64-bit support. > > It is amazing that we have to argue about that in 2014. > > Again, it is not only about about just using it but to finally have a > safe and clean 64bit implementation. Almost every other project did > that years ago, recommendation from various compilers developers, > security teams, etc. do recommend to implement 64bit support this way. > We had issues because of the lack of usage of this well known > standard. And now it is told that it is useless? I cannot agree with > this statement, sorry. > > Now, about the numbers, which are based on a prototype, there are > rooms for more improvements, we both know that. As our numbers show, > based on 5.6 and 5.6+patch, the actual impact is much smaller. And > this is without digging deeper or without looking at what would be > possible when both phpng and this patch would be merged and improved > together. We are smart people and good developers, and we both have > very good contacts to compilers expert. We can reduce the impact even > more and improve the phpng optimization as well. > > However, reducing the 64bit support to "only about allowing large > string", only to 64bit integer for long or "only about using modern > types" is wrong, and we again both know that. > > > Cheers, > Pierre > > @pierrejoye | http://www.libgd.org >