Provide a means to convert a range, possibly with unions and possibly with intersections, into something that both Excel and PhpSpreadsheet can handle. Intersections are changed into unions of the individual cells which they comprise. With this change, Xls Writer now handles intersections (previously it would have thrown an Exception or created a corrupt worksheet if this was attempted), and Xlsx Writer works correctly (it seemed to before, but Excel didn't understand what it wrote). Worksheet::getConditionalRange and ::getConditionalStyles would previously have thrown an Exception when presented with an intersection, and will no longer do so.
**NOTE:** Intersection support is limited to Conditional ranges. Use of intersections in other contexts will usually not achieve the desired result.
Fix#4039. Excel sometimes stores the location for a Conditional as, say, `B1:B10 C1:C10` rather than `B1:C10`. PhpSpreadsheet does not have a problem with this, but internally stores it as separate Conditionals, one for each range. (There may be more than 2.) User would like it stored as a single Conditional. Since the range is used as the index of an array which holds the conditionals, there is no technical reason why this can't be done. And it does seem like being able to change multiple ranges all at once has some advantages, whether you're doing it in PhpSpreadsheet or in Excel itself.
Making such a change in Xlsx Reader showed no problem in Xlsx Reader and Writer tests. A number of problems did show up in CellMatcherTest, and one in WizardFactoryTest. This was good news, because it meant some of our test spreadsheets used the same type of construction as the test spreadsheet supplied with the issue. So, if additional fixes make the test problems go away, I think no new tests are required. And a smattering of source changes did indeed result in a clean test suite once more.
This is a change, but I don't really think it should break anyone. So I don't think it's necessary to add an option to opt in to the old or new behavior. I could be wrong. I'll leave this ticket open for at least a couple of weeks to see if anyone thinks otherwise.
oleibman