Clone wiki

neurocogling / Changelog


Version 1.2.3: January 17, 2012

Version 1.2.3 is a maintenance release containing numerous fixes:

  • Fixed grid generation after resize.
  • Fixed grid viewer not always reflecting latest grid status.
  • Fixed grid saving and loading losing grid network.
  • Fixed activation gradient rendering for links of length greater than 1.
  • Fixed inhibition for links of length greater than 1.
  • Source code fixes and refactoring.

Version 1.2.2: July 17, 2011

Version 1.2.2 is a maintenance release containing several bug fixes:

  • Hopefully fixed some wierdness with the persistence functionality and stepping (#56).
  • Made arrowheads not always normal to their attach points (#57).
  • Made link labels track closer to links (#58).
  • You can now step through the network by pressing the space bar (#60).

Version 1.2.1: July 8, 2011

This version contains several bug fixes as well as three new pieces of functionality:

  • Node persistance: You can now specify a value for a node or a link that determines how many time steps the item will remain activated before it begins to decay (#53).
  • Multi-Links: A multi-link connects two grid items, and if all three items' width is the same, then activation will be transmitted precisely between them (#49).
  • Grid Viewer: A highly experimental and likely quite flaky visualization window for Grids that shows the grid's cylindrical topology in 3D. You can rotate the grid view by dragging it with your mouse (#44).

Other bug fixes:

  • Fixed the display of links to grid edge items when loading an existing grid or moving the edge item (#55).
  • Fixed the Activate/Deactivate menu options not showing up for all applicable items, as well as Reset not working properly (#52, #54).
  • Fixed a case where you could connect links to the sides of the narrow end of an Abstract OR item (#50).
  • Fixed a few typos and errors in the manual (#51).

Version 1.2.0: April 2011

This version contains one major new piece of functionality:

  • Grid Items: You can now build huge networks by building a small template network with side connections, which is then repeated many times horizontally and vertically (#11). You can experiment with text input and output using Text IO Items (#40).

Other improvements and fixes:

  • Provided a side palette with a list of network items with icons that you can drag to the editing area (#37).
  • Provided a list of recently-opened networks in the File menu (#41).
  • Vastly sped up the underlying network automaton using seqlocks instead of mutexes (#38, #39).

Version 1.1.0: November 5, 2010

This version contains two major new pieces of functionality:

  • Abstract Notation: You can now build relational networks using AND and OR nodes of the abstract (formerly "compact") notation, as well as bidirectional links (#12).
  • Subnetworks: There is a new item called a "Sub-Network", which, when you double-click on it, opens up a new network view. You can connect links to the sub-network item, and their ends will appear in the sub-network (#16).

Other improvements and fixes:

  • You can now modify the length of links (i.e. the number of time steps it will take for an activation to be transmitted along the link) (#25).
  • Fix: Oscillators can be inhibited (#28).
  • Fix: Inhibiting inhibitory links now works properly (#29).

Version 1.0.3: August 8, 2010

Bug fixes and some internal improvements.

  • Improved link rendering and made link drawing more efficient (#6).
  • Plugins are now loaded on OS X (#22).
  • OS X build system is fully automated.
  • Bug fix: fixed loading files with self-links (#23).
  • Bug fix: fixed crash when creating new item after selecting multiple items (#24).

Version 1.0.2: July 22, 2010

Mostly UI and usability improvements.

  • Added some sample files to the distribution. (#20)
  • Cut and paste: you can now cut, copy and paste network items. (#14)
  • You can now zoom in and out of the network view. (#18).
  • Printing support: you can print network diagrams or export them to a variety of formats. (#10)
  • Implemented node threshold learning similar to that in Colin Harrison's dissertation. (#21)
  • Added a network item type for text labels. (#17).
  • Added some nice icons from the Silk icon set. (#19)
  • You can now edit common properties when more than one item is selected. (#13)
  • Added a system to handle loading old file versions. (#9)
  • Made saved item type names independent of C++ name mangling. (#8)
  • Fixed a leak of automata nodes when loading and saving multiple times. (#15)
  • Misc UI efficiency and usability improvements.

Version 1.0.1

Initial release.