Release Notes Solo Predictor Version 2 8 2

Changes and Bug Fixes in Version 2.8.2

Version 2.8.2 of Solo_Predictor was released in December 2012.

Updated to be compatible with PLS_Toolbox / Solo 7.0 including support for new EVRIModel objects.

Add "plaindelimiter" configuration option to allow specification of delimiter to use in plain text output (makes parsing easer in some languages). Default is comma-separated.
Add log_backups field and add help for all backup settings
Add "header" option to include a specific header with each log file
Add json parsing and output support
Change order of "old" log files (oldest log is highest #ed log, most recent is 1)
Show which default.xml file is being used when status dialog first opens
Show statusbox message when the user attempts to close the controls using the standard window close control
Do not store debug messages in buffer if not in debug mode (avoids potential memory leak on very high-load applications)
Force message buffer dump if a large number of messages have been queued and not written
Improve status dialog performance (do not redraw menus on each display)
Better handle when 32-bit version is started without appropriate permissions for automatic configuration to operate.

Allow creation of evriscript objects directly without @evriscript(____) wrapper. Just call with: @____ and if not recogonized as a standard object but it does match an evrscript method, it will be created as an evriscript object.
Fix .mat file saving: allow -append and multiple vars
Fix for reading scripts which have a line consisting of just '%' (or '#' or '//') and a line-feed (caused following line to be skipped)
Fix for indexing into array with numerical values between 65 and 90 or 97 and 122 (would return empty or possibly throw error)
Fix to ignore leading or trailing spaces in parsed variable names or strings
Improved support for LF as end-of-command character
Add new @predictor object (limited support)
Remove support for doing 'testing' mode with | (where y-block is passed within x). New approach is to use model object directly: pred = model.apply(x,y);