Couldn't delete a value pair after certain actions (updating a parameter) until you refreshed. This was resolved.
Fixed a regression caused by yesterday's value pair fix.
Having requirements, even if they didn't have any expected results, would cause the expected results column (all empty in this case) to show up in the create tests tab. It no longer does.
Certain cases where an "any value" has no possible value were being assigned a value that violated a value pair rather than "no possible value".
File exports (Excel, CSV, zips) in IE 7 only where not completing properly. This is now resolved.
Fixed library dependency.
Window no longer has scroll bars unless they are needed.
More descriptive text to make each step clear. Made field enable/disable states and button labels more clear.
Fixed some graphical glitches in the test panel when mixed-strength tests are selected.
Email content for newly invited users now more accurately describes who invited them.
In the appropriate UI’s and export files, the parameter values that fulfill requirements or cause expected results are bolded so they can be more easily traced.
It’s easy to forget to select the higher strength tests after generating higher strength tests and then exporting. This streamlines that workflow and makes the default more sensible so it less likely to be missed.
Improved how the value of "none" is added to blank parameters on import to avoid some import failure scenarios.
This was a pair-wise fault for those keeping track at home.
For users not familiar with using numeric ranges as parameter values, all of these define valid ranges:
1,777 - 2,777,777
Spaces around the dash do no matter. So these are all fine:
1 - 5
The values for ranges are provided with the pattern: min, max, random, min, max, random, min, ....
For example, a value of "1-5" would use these values in turn in tests where the "1-5" value was used:
Random is defined as between the min and max, inclusive of them.
In the case of floating point ranges, the lesser precision of the min and max is used as the precision of the random.
Example: 1.01 - 5,5555
3.14 (random, uses 2 digits of precision, not 4)
For various reasons these DO NOT define valid ranges, and they will just return the string as the value (if you wonder what the reason is for any of them, let us know, and we can let you know):
1,00 - 2,000
1,000 - 2,00
1,000 - a
a - 2,000
1, 000 - 2,000
1,000 - 2, 000
7s7 - 888
777 - 8e8
1,00.01 - 2,000.01
1,000.01 - 2,00.01
1,000.01 - a.01
a.01 - 2,000.01
1, 000.01 - 2,000.01
1,000.01 - 2, 000.01
7s7.01 - 888.01
777.01 - 8e8.01
1.01.01 - 2.01
1.01 - 2.01.01
1.001,001 - 2.01
1.01 - 2.001,001
Negative values are not currently supported in ranges.
Switching the comma and period delimiters in the European style is not currently supported in ranges.
Thanks to Funmi for pointing out the issue.
These were related to invalid focus requests which IE treats as an error.
Parameter values in test cases that can be any value were exported as "any" in Excel and HP QC. In the HP QC and Excel exports they are now exported with the value Hexawise chose for them, and also marked with purple italics (same as in the UI) so you can recognize that these values can be replaced if needed or desired.
A work around to a prior Chrome glitch rendering the param name editor / selector was made obsolete by a Chrome fix and was now causing issues when entering long parameter names in the lastest releases of Chrome so the work around was removed.
There is a known issue in IE 9's HTML table rendering that can cause some table cells to be off by one column due to white space being treated significantly (it's not). This could be addressed by using "compatibility mode" but we made some changes in some of the HTML rendering to make this unnecessary.
Thanks to Bryan D. for bringing this to our attention.
Parameters added or edited (singly or in bulk or in import) with no defined values are now provided a single value of "none". Before the handling was inconsistent, but consisted mostly of doing nothing until the problem was corrected.
Changing selection of Excel would reset the OPML selection. This is fixed.