Hi everybody, in the past I did quite some beta-testing. The beta-testers were very helpful and proved to be a more than valuable source for improvements and bug reports. But there was one thing that bothered me... the crash-logs. You know, when an app crashes a log is generated. Inspecting the log is supposed to yield details about the origins of crashes. Unfortunately extracting logs from devices and inspecting them in XCode is quite feasible but not really convenient. The testers have to extract the logs from their devices in a multiple-step process and I have to check those logs in XCode and make sure that they are symbolicated enough to get the crash-reasons out of them. For the next game I intend to do an experiment. First of all I try to ensure that the app does not crash. This will be done by catching exceptions and putting them into an internal log. This log can be sent via email (to me) at any time. I would like to ask, do you have some insights to the issue at hand? Any anecdotes or advices?
Not sure how large scale/intimate your beta group is, but my experience has always been that only a handful of people will go to the trouble of emailing you anything/going online for a survey, etc. A better approach, is to just push that data out to a database from your beta builds. That way you can push out errors as well as important gameplay details: like, how long on average does it take a player to get through level 2. We just informed out beta testers that these builds actively push data to the net (as well as check if the current beta build is out of date and notify them) and mined the data as it came in. Useful for error catching and super useful for gameplay balancing/details.
coby_pxs, your idea is more than excellent! Thanks a lot! I think a good way would be this: 1. send logs to a (My)SQL-database 2. if there is no internet connection, buffer the log entries that you want to send for a later retry 3. use some username to distinguish the log-entries 4. make use of some independent evaluation mechanism I believe this could be done in no time. Of course it is crucial that the beta-testers agree to that policy. If implemented, agreeing to that policy should be a requirement for beta-testers. Any more thoughts on that?
