As we're having fun with StyledText and editors these days, some of us contributors have had discussions about how to test things.
I'd like to share with you a simple and powerful approach to test Text rendering (in a widget, in an editor, wherever): using an OCR like Tesseract.
I've tried the following:
* Open a Java editor with codeminings enabled
* Took a screenshot of the editor (cropped to remove ruler and keep only text)
* then ran from CLI `tesseract /path/to/screenshot.png stdout -psm 6` and got as output a string that matches almost perfectly the text in the editor; more particularly showing the codeminings were rendered and read properly.
I think it's an extremely powerful way to do rendering test for StyledText and the family (editors, code minings...).
Would is seem acceptable to start adopting such OCR-based tests where needed in Platform? I really think it's 1000 times worth it.