If a Scrum "increment" is to be "done" by the end of the "sprint", where "done" means usable/releaseable, I assume testing has to be part of the sprint. Since sprints are not supposed to last more than a month, depending on the scope of the sprint goals, testing is either ongoing or occurs just before the end of the sprint. It seems on-going has to be the answer, but regressive testing of the new increment with previous incrments can grow into an increasing work item.
Does anyone with experience in this have thoughts to share so I can understand better how this is done?