tl;dr Mike is an amazing tester and found a complicated latent defect in Hexawise that had gone undiscovered for years. Unfortunately there's no simple way to describe this defect, so our apologies for the complex prose if you chose to read on...
If you have a large number of requirements and they cause 1-way coverage (all values of the parameter) to be completed on a parameter with a large number of values (let's call it parameter A), and you still have additional requirements after completing the 1-way coverage on parameter A, but those requirements didn't also require a value for the parameter A, you could trigger a defect where Hexawise would use a value of parameter A in a test that was satisfying a requirement to achieve 2-way coverage with a value of another parameter, without also locking in that value of A as that value of A for that particular test. The "any value" that is then reported for A in that test is not accurate, as a specific value is required there, and more problematic, the "any value" replacement value eventually provided for value for A might not be chosen as the same value that was used to achieve the 2-way coverage, leaving your generated tests without complete coverage.
This unusual set of circumstances is the first defect we've ever found in Hexawise that results in incomplete coverage. Major kudos to Mike for both finding the defect, providing a brilliant defect report, and also doing some initial detective work that made isolating the bug simpler. We have some pretty amazing testers as users of Hexawise!
This only occurred AFTER canceling a valid lower strength test case generation and in cases where the higher strength test plan was too large (or invalidly reported as too large).
A fairly classic 3-way defect. This one had been in production for a while without being reported.
The strength selector drop down was providing invalid warnings about large overly large generated test sets in some cases.
Thank you to Tyler for reporting the issue.
If you share a project, AND the plans in the project have requirements, AND you share the project via the sharing URL (each sharee gets their own copy), AND you modify or delete the parameters referenced in the requirements back in the original plan in the original project, then the requirements would be blank.
Count up those "AND"s above and you'll see we have a classic 3 to 4-way defect depending on exactly how you identify the variations. It was found in production by a user (defect has been around for a year or two).
Thanks to Steve for finding the issue!
The test description field template was limited to 255 chars, which was insufficient for some uses. This has been vastly expanded.
Hexawise attempts to follow the logical implications of the value pair constraints (invalid and married pairs) you create and automatically creates many of the value pairs that are implied.
There is now a visual distinction. Value pairs you created are normal, while the implied value pairs Hexawise created are semi-transparent.
Thanks to Steve for the feature suggestion.
In some cases value pairs would not highlight on hover in the define inputs page immediately after they were created until there was a page refresh.
The spacing and appearance of this dialog has been improved to be more readable.
When exporting from the requirements page, you are notified if there are unsaved edits to the requirements that will not be included in your export.
When exporting from the auto-scripts page, you are notified if there are unsaved edits to the requirements that will not be included in your export.
Thanks to Kevin for the feature request.
Which parameter value you click on first when making a unidirectional married pair is significant as it determines the order of the constraint. "Every chimp is an ape", is not the same as, "every ape is a chimp".
The fact that the order is significant was not the most discoverable feature in Hexawise so we "promoted" it into the dialog where you select if the married pair is bidirectional or unidirectional. You now also select if the unidirectional married pair is reversed or not. The order that you click the parameter value in is now only significant for which selection is defaulted.
This is fixed now.
In certain instances an error would occur trying to cancel computation of a plan resulting in no cancellation of the request.
The link in the plan import dialog to a list of mind mapping tools that support OPML was broken.
Hexawise no longer supports IE 7, which was released on October 18, 2006.
Hexawise now officially supports Internet Explorer 10 and 11.
Test strengths higher than the number of parameters are no longer shown in the strength selector.
The wording of the tool tips on the coverage chart are now clearer.
If all you changed was a parameter value by clicking on it to edit it the plan cache was not cleared and the prior test results were retained until something else changed. Thanks to Anita and Kevin for reporting the issue.
There were some avoidable strictness issues with blank Excel sheets and unmerged cells that the plan importer could be more forgiving about and deal with. Now it does.
Thanks to Jafar for reporting the issue.
You can now click on an individual parameter value name on the Define Inputs page to edit it.
Editing a parameter value name individually preserves it in places where it's referenced rather than deleting those references:
Thanks to Jafar for reporting the issue.
Before you had to guess. Now... not so much.