| … | … | @@ -102,6 +102,14 @@ reads and writes are described by this graph: an AtomSpace accessible |
| 102 | 102 | for reading is also accessible for writing (unless it is marked |
| 103 | 103 | read-only.) |
| 104 | 104 | |
| 105 | The current implementation of layered AtomSpaces has some strong |
| 106 | negative performance reprecussions. This is because the layers are |
| 107 | envisioned as being OverlayFS-like: Atoms in lower layers can be masked |
| 108 | or hidden in higher layers, so that it looks like they are absent. |
| 109 | Atoms in higher layers can have different key-value store contents |
| 110 | than lower layers. The lookup has fairly strongly noticable performance |
| 111 | impacts, even though this is not benchmarked at this time. |
| 112 | |
| 105 | 113 | This directed graph is represented with a Frame, which is an Atom that |
| 106 | 114 | is both like a Node and like a Link, having both a (singular) name and |
| 107 | 115 | an outgoing set. |
| … | … | @@ -167,8 +175,14 @@ take a closer look and make sure we are not missing anything. |
| 167 | 175 | multiple parties can engage in coopertive actions by authenticating |
| 168 | 176 | with crypto keys. |
| 169 | 177 | |
| 170 | | How can we leverage the above for the next-gen design? The |
| 171 | | implementation of AtomSpace layers |
| 178 | How can we leverage the above for the next-gen design? |
| 179 | * The implementation of AtomSpace layers was challenging. |
| 180 | * The current implementation has an underwhelming performance profile. |
| 181 | * The current ProxyNode implementation is already version 2; |
| 182 | the version 1 variant had proxies, but not ProxyNodes. |
| 183 | |
| 184 | So we're effectively going for version 3, here; lets get it right, |
| 185 | instead of being half-assed about it. |
| 172 | 186 | |
| 173 | 187 | |
| 174 | 188 | ### Are SensoryNode's AtomSpaces? |
| … | … | @@ -186,3 +200,6 @@ TODO List |
| 186 | 200 | --------- |
| 187 | 201 | Writiing the above, the TODO list expands: |
| 188 | 202 | * The ThreadJoinLink should be removed from Atomese. DONE. |
| 203 | * Benchmark the Atompsace layers/frame performance. Maybe with a variant |
| 204 | of the `LargeZipfUTest` layout, but this time with different count |
| 205 | data at each layer. |