Changes between Version 12 and Version 13 of soc/2007/SignalNetwork

Timestamp:

Jul 5, 2007, 8:50:29 PM (15 years ago)

Author:

Stjepan Rajko

Comment:

--

soc/2007/SignalNetwork

@@ -1 +1 @@
 == Prototype implementation ==
 
-If you're interested, you may want to examine the [http://dancinghacker.com/code/signet/ documentation for the prototype implementation].
+If you're interested, you may want to examine the [http://dancinghacker.com/code/signet/ documentation for the prototype implementation] (the documentation only reflects changes up to 2007/06/14).
 
 == News ==
-'''(06/14/2007)'''
+'''(2007/07/05)'''
+ * everything is now in boost::signals namespace
+ * directory structure is now more modular
+ * fused/unfused is now a template parameter
+ * connection mechanism is more easily extensible and modular
+ * all components can now accept both fused and unfused signals
+ * storage has been reimplemented through a generic component
+ * storage send signal is now separate from operator().
+ * test_disconnect now fails?
+
+'''(2007/06/14)'''
  * added branching tests as separate files
  * finished updating docs for Boost community update
 
-'''(06/12/2007)'''
+'''(2007/06/12)'''
  * Implemented generic applicator, modifier, conditional, and instantiator components which do various generic things to signals.
  * Some of the old components have been converted to derive the generic components.
@@ -14 +24 @@
  * Tests have been split up into separate files.
 
-'''(06/03/2007)'''
+'''(2007/06/03)'''
  * I have reimplemented all of the prototype components (except for selector, which needs to be redisigned) using Boost.Fusion.
    Unfused versions of the components are made possible by two classes which I have adapted from fusion's unfused_typed class
@@ -26 +36 @@
 == Proposed changes ==
 
- * Through trying different approaches, the fused / unfused component selection mechanism is implemented awkwardly and is nod ideal to use.  I'd like to make fused vs. unfused to be decided by a template arg as is done with signet::filter, rather than as component vs. component::unfused
- * The rest of non-generic components need to be converted to derive the generic components
- * Revisit the naming / namespaces... Probably make a separate namespace and include file for connection operators, in case they are not needed.
+ * I have started to experiment with non-signal based data transport mechanism - this library will probably be renamed to dataflow and offer both mechanism-independent dataflow functionality, as well as specialized modules for specific mechanisms (like signals)
 
 == Introduction ==