Thu, 23 Oct 2008 17:47:46 +0200, Don wrote: > Sergey Gromov wrote: > > Wed, 22 Oct 2008 15:56:30 +0200, > > Lars Ivar Igesund wrote: > >> In any case, a hierarchy of the type > >> > >> common/ > >> core/ > >> sys/ > >> stdc/ > >> > >> should be highly considered. This would allow a namespace for functionality > >> that is truly common, not only the runtime, but math and eventually other > >> functionality. In addition it is naive to believe that just because > >> druntime is meant to be a common runtime, that it will be the only runtime > >> in the long run. > > > > There is no 'truly common' functionality beyond the absolutely necessary > > core runtime. The fact that Phobos and Tango share some code only means > > that such code is ought to be in any stand-alone, general-purpose > > library. Not that it must be built-in. > > ??? Don't understand. > Are you simply saying that there's no need for standard libraries?
I can see a reason for a general-purpose library which is supplied with a compiler. It's for faster setup, faster learning, and for quick-n-dirty utilities with less dependencies. I can see why one would want a different general-purpose library instead of the default. It's for more appealing (for that one person) library architecture and for different trade-offs. I can see a reason for a specialized math library, biological, chemical library etc. They're specialized. But I cannot justify a separate library only because Phobos and Tango happen to use the same approach in some parts. A library without any architecture, with only purpose to make Phobos and Tango dependent upon it, and to make users wonder where to search for a particular general- purpose functionality.
