Having now used Sentry in production for over a year, our ultra-#1 feature request would be the ability to update an event after the fact, especially upload files – AFTER we’ve asked users what we’re allowed to upload.
When our app crashes, we’re currently forced to immediately decide what gets uploaded. This is impossible. We’d like to upload the user’s document. We’d like to upload screenshots, logs, lots of excellent stuff for troubleshooting. But we have to ask.
Sentry makes this basically impossible, because it’s based on the idea that the entire decision about what to upload can be made passively. This is completely naive. The most useful information for debugging is also very privileged, and we’d be breaking countless laws and moral standards if we simply uploaded it. The dance we have to do in order to postpone the crash upload and manipulate the suspended crash upload is completely crazy and unreliable.
Sentry should live in a world where you can ask the user what to submit with the crash report. And I think that rather than postpone the crash report until the user has answered, it would be best to be allowed to augment the original event after the fact.
Simply being able to submit user feedback is not terribly useful. User feedback usually sounds like “something happened somewhere”. But when people send their files, or screenshots, we can nail bugs very quickly.
We’ve somehow managed to work around this deficiency for our C++ project, but for our C# project, it’s becoming impossible to do anything, because we don’t have the low level access to hack around Sentry as we do in C++. And we kind of resent having to jump through hoops to submit information in the proper order in C++.
So the only feature request that really matters to us is being able to upload information to an existing event, such as files or tags, AFTER we’ve asked the user what we’re allowed to include.