Difference between revisions of "Analytica 4.0 beta Release Notes"

(Moved 4.0.0.49 to its own page)
Line 27: Line 27:
 
= Changes in 4.0.0.50 =
 
= Changes in 4.0.0.50 =
  
 +
Build 50 is solely a bug-fix release.  While I don't usually list minor or straight bug fixes in the release notes, I will do so here.  The first bug listed is the most significant fix in this build.
 +
 +
== Edit Tables ==
 +
 +
* Fixed a bug present in build 4.0.0.49 which would cause large edit tables to break when a cell was changed.  This was a pretty serious bug, such that anyone using 4.0.0.49 definitely should upgrade to 4.0.0.50.
 +
 +
== Uncertainty Settings ==
 +
 +
* Fixed the "reset once" option on the uncertain dialog, which was not invalidating previous samples & computations as it should.
 +
 +
== Evaluation ==
 +
 +
* [[DatePart]](..,"w") fixed to return 1=Sunday, 2=Monday, etc., as specified on the Wiki page.
 +
 +
* An error message is now issued if a result for a time-slice in [[Dynamic]] contains an unindexed list (aka "implicit index", "null index").  An invariant enforced by Analytica is that any array can be dimensioned by a given index at most once, and as an extension of this rule, the implicit index can only be there at most once.  In a dynamic loop, if an implicit index results, this creates an ambiguity akin to having multiple implicit indexes in the same result.  For example, consider: Dynamic([1,2], [3,4,5]).  The final result would have to have two implicit indexes.  Even though you might argue that Dynamic( 1, [2,3] ) would not be ambiguous -- since the same [2,3] would appear in all slices beyond the first and could be assumed to be the same implicit dimension, we've still signalled this as disallowed in this build.  In former releases, this was not signalled as an error, but problems results when this condition occurred, and thus we doubt it was employed.  If you think you need a list as a dynamic element, you should be able to utilize a local index, see [[Index..Do]].
 +
 +
* An obscure bug occurred when the sequence to a FOR loop was an unindex list passed in as a passed in as a parameter to a user-defined function.  The circumstances for this bug to occur are rare, but several possible manifestations might occur:  An "internal type mismatch error", a stray "Index  has not been explicitly defined" error, a crash, and/or a stray blank index in the result appearing just as a dot.
 +
 +
== Saving ==
 +
 +
* When an index element (a label) contained a lone [[Chr]](10) character (this would only occur if the label itself is computed), and the model was saved while the index was in an [[Table Splicing|unspliced state]], the model file would be in an invalid format that could not be reloaded (continuation tildes preceding the Chr(10) are necessary, but were not output).
  
 
= Previous Beta Builds =
 
= Previous Beta Builds =

Revision as of 17:50, 16 July 2007

(Back to Beta Tester Page)

These Release Notes apply to Build 4.0.0.50, 13 July 2007.

Installation Notes

  • Analytica 4.0 can coexist with Analytica 3.1 on the same computer. Thus, you can install Analytica 4.0 beta while retaining Analytica 3.1. After installing 4.0, a double-click on a x.ANA file will run the model in 4.0. To run a model with 3.1, start Analytica 3.1, and use the Open Model... option from the File menu to select the model.
  • If you uninstall Analytica 3.0 or 3.1 after installing 4.0, it will wipe out the 4.0 registry tree. So, you will need to re-run the 4.0 installer.

Upward compatibility for Models from Analytica 3.1

Most models created in Analyica 3.1 load and evaluate fine in 4.0. We hope beta testers will check that 4.0 works and gives the same results as 3.1 -- unless there is a good reason why they should be different. There are two areas where you should expect results to be different:

  • Graph settings: Some Graph Setup options will not transfer to 4.0, because 4.0 offers different (and much richer) graphing options. For example, settings for axis ranges in 4.0 may attach to the Index rather than the graph. So, expect to do some tweaking of graphs -- and to enjoy the new options.
  • 0*NaN and 0*INF: Analytica 4.0 supports the IEEE 754 Binary Floating Point Arithmetic standard: Thus 0*NaN or 0*INF return NaN. In 3.1, the result was zero. If your model took advantage of the old behavior to zero out NaNs or INFs, you may find NaN or Null in downstream results that were not there in 3.1. If you check Show result warnings in the Preference dialog, it will warn you if your model relies on the old behavior and help pinpoint where this occurs.

See Changes in 4.0 that could impact 3.x models for a more comprehensive list.

Not working yet

  • There may be some problems saving models in XML format, especially related to system variable settings. For important stuff, stick with the default Analytica file format for now.
  • Auto recovery does not correctly recover "Add Linked Module" actions. Some problems after "Add Module". OLE links are not re-established. In the latter two cases, saving the model after crash recovery, then reloading should fix the problem. It is best to save after adding a module or library, rather than relying on auto recovery.

Changes in 4.0.0.50

Build 50 is solely a bug-fix release. While I don't usually list minor or straight bug fixes in the release notes, I will do so here. The first bug listed is the most significant fix in this build.

Edit Tables

  • Fixed a bug present in build 4.0.0.49 which would cause large edit tables to break when a cell was changed. This was a pretty serious bug, such that anyone using 4.0.0.49 definitely should upgrade to 4.0.0.50.

Uncertainty Settings

  • Fixed the "reset once" option on the uncertain dialog, which was not invalidating previous samples & computations as it should.

Evaluation

  • DatePart(..,"w") fixed to return 1=Sunday, 2=Monday, etc., as specified on the Wiki page.
  • An error message is now issued if a result for a time-slice in Dynamic contains an unindexed list (aka "implicit index", "null index"). An invariant enforced by Analytica is that any array can be dimensioned by a given index at most once, and as an extension of this rule, the implicit index can only be there at most once. In a dynamic loop, if an implicit index results, this creates an ambiguity akin to having multiple implicit indexes in the same result. For example, consider: Dynamic([1,2], [3,4,5]). The final result would have to have two implicit indexes. Even though you might argue that Dynamic( 1, [2,3] ) would not be ambiguous -- since the same [2,3] would appear in all slices beyond the first and could be assumed to be the same implicit dimension, we've still signalled this as disallowed in this build. In former releases, this was not signalled as an error, but problems results when this condition occurred, and thus we doubt it was employed. If you think you need a list as a dynamic element, you should be able to utilize a local index, see Index..Do.
  • An obscure bug occurred when the sequence to a FOR loop was an unindex list passed in as a passed in as a parameter to a user-defined function. The circumstances for this bug to occur are rare, but several possible manifestations might occur: An "internal type mismatch error", a stray "Index has not been explicitly defined" error, a crash, and/or a stray blank index in the result appearing just as a dot.

Saving

  • When an index element (a label) contained a lone Chr(10) character (this would only occur if the label itself is computed), and the model was saved while the index was in an unspliced state, the model file would be in an invalid format that could not be reloaded (continuation tildes preceding the Chr(10) are necessary, but were not output).

Previous Beta Builds

Comments


You are not allowed to post comments.