Analytica 4.0 beta Release Notes/4.0.0.42

Changes in 4.0.0.42

(since 4.0.0.39)

This section is in-progress. Refer to the links at the bottom for notes on previously released beta builds.

In general, bug fixes (and there are many since 4.0.0.39) are not listed here.

Standard attributes should now always show in the attribute panel dropdown at the bottom of a diagram, regardless of whether they are checked in the Attributes.. dialog. As before, User-defined attributes are there only if checked in the attributes dialog.

A couple menu items have been renamed.
- Window->Tile Horizontally is now "Tile Top to Bottom"
- Window->Tile Vertically is now "Tile Left to Right"
- Diagram->Align->Horiz. Centers is now "Centers Left and Right"
- Diagram->Align->Vert. Centers is now "Centers Up and Down"

Some new shortcut keys have been introduced.
- Function keys, F2-F9, correspond to the toolbar buttons (left-to-right). e.g., F2=parent diagram, F3=outliner, .., F9=arrow edit mode. F10 remains a key to access the main menu. F11 toggles between edit/arrow modes (formerly F2). And F12 or CTRL F12 provides access to the typescript window (same as CTRL-', but the latter is not available on some non-English keyboards).

Find... will find exact matches first. For example, when searching for a title, if you have a node titled "Revenue" and another titled "Net Revenue", a title search on Revenue will find both (as before), but will find the exact match first. As before, CTRL-G finds the remaining matches.

When you enter expressions into the text boxes in the object finder for an operator, the function finder is now aware of operator precedence and will insert parentheses when necessary. For example, if you enter "1+2" and "3+4" into the boxes for the * operator, the expression (1+2)*(3+4) is pasted into the definition.

New alignment option: Align->Left and Right Edges
Make Same Size->Width (or Both), when applied to input nodes, now resizes the input control, as well as the node boundaries.
Align->Horiz. Center (Centers Left to Right), when applied to an input/output control, now uses the left edge of the control as the point of alignment.
Make Same Size, when applied to an input/output control, always leaves the left edge position fixed (rather than leaving the center fixed as with variable nodes).
When you have more than one variable selected when you select "Show result", a new node is introduced. This comparison node is now initially named, e.g., "Compare1", and is defined as a list of identifiers rather than a self-indexed table as it previously was.
Align, Make Same Size, and Space Evenly are now all undoable.

Choice pulldowns in edit tables have been shortened and widened for asthetic reasons, to reduce unnecessary row height and to fill out the cell.

Tables downstream of ODBC-read data, or other forms of exogenously-determined data, can now auto-splice when a model is read and the exogenous data has changed since the model was previously in memory.

The Insert Cells and Delete Cells operations are undoable, even when many tables are spliced as a result of the operation (undoing recovers data in all altered tables).

The default graph background was changed back to white. (In previous beta-builds, it was set to a light blue gradient).

Two pre-defined graph templates have been added. Beta testers should not rely on these, they are subject to change. But we do hope to assemble several pre-defined templates before release, so these have been included so we can work out the kink with pre-defined templates.

When there is more than one key on a graph, the key background (when outlined or filled) is now a single rectangular area that covers all keys. Previously each key was outlined separately.

Pivot controls are positioned in two columns, if the window is sufficiently wide to allow this without overlap. They are also lined up more systematically.

When a dimension contains variable handles, either the title or identifier will appear on the graph (for example, as axis labels), depending on the "Show By Identifier" (CTRL-Y) toggle.

The maximum number of user-defined attributes (UDAs) is now more than 100. Previously only 5 UDAs could be defined. However, the limitation of showing only the first 5 on the object window remains. If you need to view/change UDA values for the 6th or greater UDA, use the attribute pane or typescript.

Analytica 4.0 has been moving away from expressions that return the special value Undefined. Result with no result in 4.0 will all return Null consistently. A few additional updates towards this goal have occurred in this build. If-Then without Else, MdTable's default for missing values.

We are moving away from the term "VarTerm" for Meta-Inference, in favor of the term Handle. Function Handle( ) was introduced to replace Function VarTerm (which remains, but is not exposed). Also, HandleFromIdentifier now replaces GetVariableByName (the latter is still there, but again, not exposed).

The constraintsOnly parameter has been removed from LpFindIIS and LpWriteIIS. This had been added in 4.0, but we decided it was unnecessary.

The CumDist distribution function was extended to support named parameters, to be usable within Random, and to support the Over parameter. In addition, a new optional parameter, smooth, has been added, which controls the interpolation method.

ProbDist was extended to support named parameters, to be usable with Random, and to support the Over parameter.

There is now an option to always show the file dialog when calling ReadTextFile and WriteTextFile.

A new statistical function, Covariance(X,Y), has been introduced, for estimating the sample covariance (or weighted covariance) from a sample or from data.

Text functions automatically coerce to text without complaining if their parameter is not already text.

The Multivariate Distributions library has been extended substantially. All multivariate distributions support the Over parameter and can be used with Random. Several functions for generating serially correlated distributions are available. And RegressionDist provides a generalization of the Regression function, which returns a distribution reflecting parameter uncertainty due to finite data.

New example added: "Disease Establishment.ana". (not related to new 4.0 features)