I guess duty calls.
I'm a software engineer with a large multinational computer systems company, and I've been interested to see how Spotify handled the first significant bug I've seen in their product since I first started using it around 5 years ago. In short, they did about as well as could reasonably expected of them, though perhaps they could have communicated better so that some of the more excitable users would have understood that the problem was actually being worked on.
It's important we understand that is not a simple bug to reproduce, because it required some unknown amount of time to pass while context-switched to some other program for it to occur (somewhere between 20 and 45 minutes in my experience). That's a long time for a computer, and any number of things could have happened during it to trigger the crash. It's also a long time for a developer to wait to reproduce a bug (about a 10th of their working day), and because of the uncertainty around what during that time was triggering the issue, it makes it difficult to create an automated test for the crash. As a result, just causing the bug and capturing the necessary information about the programs state and context was harder for this bug than most of the others they deal with. It's exactly this kind of bug I dread having to deal with.
They did request the .dmp crash files and console tracing in this thread (or one of the many that ended up merged into each other), but the process for collecting them and submitting them wasn't really obvious from the posts, which I think confused quite a few people. One way they could improve here would be to automatically collect the crash files when spotify restarts, or to otherwise streamline their process for submitting the data.
I started experiencing this issue about two weeks ago. I think it took about a week for enough other users to report the issue that the moderators took notice and started escalating to support. From there it took about another week for the developers to reproduce and analyse the issue, develop and test a fix, integrate the fix with the product and roll out a new release to the user base. Frankly, for an issue that seemingly only affected some users if they performed a specific action in the desktop client only, that sort of turn around is actually pretty remarkable.
Yes, these things are annoying when they happen, but nobody's life depends on deleting tracks from their playlists, and considering all the functionality they do offer which doesn't have any obvious flaws, I think they're doing a pretty decent job. The company I work for will only provide fixes for similar issues as quickly as this if the customers experiencing it are similarly large multinationals who are likely to be losing money due to the defect.
About the only improvement Spotify could reasonably make in the future is to communicate better about what stage the bug is in: triage/analysis/development/testing/integration or however they slice things up. That and making it easier for users experiencing a crash to submit the necessary core files and tracing.
Thanks Spotify 🙂
PS: While I'm at it, Spotify has over 6 million paying subscribers, so they're not going to notice your protest boycott. I'm sure they'd rather rdio had to deal with your whining 🙂
Staff / Moderator/ 2 years ago in Social & Random