From: Morris, C. <chr...@sn...> - 2001-12-17 13:44:58
|
> > We need categories. We need to be able to run individual > ones, a complete > > category, or all. We only have a couple of dozen tests, but > soon we will > > have hundreds if not thousands. > > If you use only one-level-deep test suites, it will be very > much like Boxster's categories. Yeup. I guess the tree view could be a hassle if I was browsing for a specific test. Once I expand a category node in a tree view, I lose sight of the other categories. A simple collapse and I'm back, but with the Boxster GUI I could easily browse the categories and see the tests in the list view. It saves a step. > > > The treeview "looked" cool, but it would slow down our > > navigation. > > I don't understand why. TreeViews can be searched just by typing. After seeing the Boxster GUI ... it's a good GUI, but I don't see how it's better at navigating to a specific test, especially once you have the 100s-1000s of tests you'll say you have. If I have 100 tests in one category in the Boxster GUI, I click the category, then I have a list box of 100 tests to look through. DUnit GUI, expand the category node and have 100 child nodes to scroll through. Neither's groovy in that case ... both list box and tree view allow you to travel faster by typing the first characters of the test name. It could be handy to have a GUI filter, show me only test nodes that match this pattern, etc. One thing the Boxster GUI seems to lack is persisting a group of selected individual tests between runs. This is how I worked with adding new tests to the suite with DUnit: When I'm writing those tests and that code I only want to run those tests. When they're first added to a large suite hierarchy, I do need to do some navigation to get at it, but then I select only those tests I want to run and that selection stays put next execution time, so I don't really need fancier navigation. Once I'm done with a test and its production code, I rarely need to execute that test again unless it fails, and there DUnit helps me by allowing me to select all failed tests in one shot. But I suppose if a test suite got *really* big, you could just organize your code such that you can run one master suite or just run sub-suites themselves as you work with it. Then when that detail work is finished, re-run the master suite. Chris |