The C++ 11 standard happened a few years back, and the direction the language has taken profoundly changed quite a bit about programming for me. The language has become much more expressive, efficient and a lot more fun to program with.
As I have become more familiar with the library features present in C++ 11 and C++ 14, I have started to see some areas where the cross-platform portability of the standard library is deficient for systems-level programming, or where the C++ library really needs some supplemental functionality.
This project is a collection of reusable and hopefully library-grade code that supplements the C++ 11, 14 and latter standards.
There are several parts of the C++ standard library that I think could use a complete overhaul. The most glaring one is the iostreams framework. I personally am still very unhappy with the speed and interface C++ provides for doing just basic terminal and file I/O.
Additionally, I think there's a lot of work that can be done to provide useful abstractions on top of file descriptors, and provide a natural Posix-like interface.
In coming to understand the above issues, a third area for improvement has emerged as something that needs a reimagining, and that area is localization.
Boost, by design, is a set of cross-platform libraries. XOS is not meant to be portable in the same sense. Everything here is meant to be as thin of a wrapper as possible around OS facilities. The rationale behind that is that it allows for better abstraction of the underlying OS.
Perhaps some code here will end up in Boost, perhaps some code from Boost will end up here (or get re-implemented for licensing compatibility reasons). That isn't a primary goal, however.
As of December 2014, there's a number of useful library pieces here that can be used to simplify some common programming tasks. The development of components of this project is following a willy-nilly 'whatever I find useful to add' approach.
There's a lot of code that is in varying states of completion and maturity; below is a list of the components that feel more mature to me:
-
src/codecvt - This is an implementation of the missing std::codecvt<char16_t> and std::codecvt<char32_t> class specializations for the gcc standard library (libstdc++), along with the missing codecvt_utf8, codecvt_utf16 and codecvt_utf8_utf16 additions for C++11/14.
-
src/filesystem - This is a near complete implementation of the std::filesystem proposal, as detailed in N4100.
-
src/environment - This is an object-oriented replacement for GNU getopt and getopt_long. It is still very new code, but is based on something I wrote and have been copying into many new programs.
-
src/time - A small collection of functions that map the low-level Posix clock_gettime calls in an std::chrono::clock style interface, and functions to convert std::chrono::time_point and std::chrono::duration to and from struct timespec's. This code was originally written by me to work around older versions of GCC not having a working std::chrono::high_resolution_clock, so it is less useful than it once was, but still nice to have an interface to std::chrono that allows the direct specification of which system clock you wish to refer to.
-
src/error - Contains some code that can be linked into an application to provide a stack trace of from where an exception was thrown. This is not meant to be used by a regular application, but to be linked in as an aid in debugging where an exception originated.
-
src/utility/average.h - A class to accummulate an average, standard deviation and variance from data, designed to provide O(1) sampling and average computation.
-
src/utility/bithacks.h - A collection of bit manipulation functions that leverage compiler intrinsics and inline-asm to provide a nice interface to use from C++ programs. Much of the code here has been adapted from the book 'Hacker's Delight'.
-
src/utility/util.h - A lose federation of very useful functions for string manipulation, exception generation, and RAII wrapping of C library functions.
The current status is that very little work on the loftier goals of the project has been started. As with any software project, you need to start out with a firm foundation on which to build abstractions.
Because I have limited time to work on this, the primary platform I am targeting is x86_64 Linux. I am not interested in building or testing this code on other platforms, though i do welcome contributions from people that do have an interest in using this on other Hardware/OS combinations.
As time goes on, I believe that this project will become very Linux-centric. As I start to tackle some of the file descriptor abstraction, I expect the epoll interface to become very prominent in the code.
Right now, the focus is to offer a nearly complete implementation of the filesystem draft proposal (n4100).
I chose XOS for the name as a 3-letter name that is short for Linux/Unix/Posix-OS.
