There was only a warning for invalid pairs. Thanks to Kevin for reporting this.
In some rare cases, the check that prevents creating constraints in some of the cases that will cause no possible values would cause a system error.
Hexawise can now detect 3 additional causes of "no possible value" in generated test cases and proactively prevents them when you are adding value pair constraints.
As a performance optimization, Hexawise will leave "any value" as the value in some test cases of large plans that are heavily constrained rather than suggesting a value in purple italics that will satisfy all constraints.
There was an issue where some simple plans would incorrectly report themselves as complex plans and trigger this optimization.
Thanks to Kevin for reporting this issue.
If you create a new test plan and go right to auto-scripts after adding some inputs without first generating tests, the auto-scripts editing would be broken due to the auto-scripts preview functionality not being able to initiate.
The order and sequencing of user actions is an important source of variation that is often overlooked but can be the source of numerous defects. Keep this type of variation in mind when designing your test plans.
The suggestion for reducing the number of test cases by reducing the largest parameters is now more nuanced as to when and how it is applied.
The Test Plan Scorecard now maxes out at 100% when reporting the percentage of parameter values that are constrained. It doesn't make sense for this one to go all the way to 11.
When plan computations max out at less than 6-way, Hexawise now explains why the higher strength test computations are not available and the very large number of test cases that would otherwise result.
Thanks to Robin for reporting this issue.
Any value replacement has to check for value pair constraints and so is inefficient as the numbers get very large. We updated the heuristics we use to decide when to leave "any value" without a replacement.
A confirmation dialog is shown so you know the operation completed successfully, and the new project is scrolled into view in the plans/projects panel.
A sequence of every possible allowed structural variation in an OPML import file would confuse the importer on some parameters.
This is a classic pairwise defect. Our tests had the variations in isolation and in combination, but not at a strength (3-way) required to trigger this bug. Embarrassingly our own sample OPML input file would trigger this bug since it was an exhaustive example. (We've since added the sample to the test suite).
Thanks to Kevin for reporting this issue.
It's often a sign the test plan may be too large in scope and would be better as multiple test plans.
In Hexawise equivalence classes are expressed with value expansions. Since they are considered equivalent to each other for the sake of the test plan, every value expansion value is not guaranteed to appear in at least one generated test case (which is why they can be a key strategy to reducing the number of test cases). The plan scorecard now lets you know when you have value expansion values that do not appear in any generated test case so you can make an informed decision about the trade offs.
Handy popovers.
Pasting email address now enables the reset button without typing. This is a 2-way fault of a parameter that's often overlooked. Was the value typed or pasted?
You'll now receive an email when it's time to retry a quiz.
Not so cramped.
Oooh... pretty.
Just a bit easier to read and use now.
Thanks to Remin and Courtney for reporting this.
A rare 3-way fault in production! Now fixed.
There is now a set of observations and recommendations about your test plan available in the "Analyze Tests" tab, and in a sheet in the Excel export.