Added more comments, help, and socket actually reads and writes some, but it's

not done.  I need to decide how I want to do the buffering...

1 files changed, 47 insertions, 0 deletions

diff --git a/src/archive.h b/src/archive.h
index 9ac3303..2f782d3 100644
--- a/src/archive.h
+++ b/src/archive.h
@@ -9,6 +9,53 @@
 
 namespace Bu
 {
+        /**
+         * Provides a framework for serialization of objects and primitives.  The
+         * archive will handle any basic primitive, a few special types, like char *
+         * strings, as well as STL classes and anything that inherits from the
+         * Archival class.  Each Archive operates on a Stream, so you can send the
+         * data using an Archive almost anywhere.
+         *
+         * In order to use an Archive to store something to a file, try something
+         * like:
+         *@code
+     * File sOut("output", "wb"); // This is a stream subclass
+         * Archive ar( sOut, Archive::save );
+         * ar << myClass;
+         @endcode
+         * In this example myClass is any class that inherits from Archival.  When
+         * the storage operator is called, the Archival::archive() function in the
+         * myClass object is called with a reference to the Archive.  This can be
+         * handled in one of two ways:
+         *@code
+         * void MyClass::archive( Archive &ar )
+         * {
+         *      ar && sName && nAge && sJob;
+         * }
+         @endcode
+         * Here we don't worry about weather we're loading or saving by using the
+         * smart && operator.  This allows us to write very consistent, very simple
+         * archive functions that really do a lot of work.  If we wanted to do
+         * something different in the case of loading or saving we would do:
+         *@code
+         * void MyClass::archive( Archive &ar )
+         * {
+         *      if( ar.isLoading() )
+         *      {
+         *              ar >> sName >> nAge >> sJob;
+         *      } else
+         *      {
+         *              ar << sName << nAge << sJob;
+         *      }
+         * }
+         @endcode
+         * Archive currently does not provide facility to make fully portable
+         * archives.  For example, it will not convert between endianness for you,
+         * nor will it take into account differences between primitive sizes on
+         * different platforms.  This, at the moment, is up to the user to ensure.
+         * One way of dealing with the latter problem is to make sure and use
+         * explicit primitive types from the stdint.h header, i.e. int32_t.
+         */
         class Archive
         {
         private: