October 2022 update
Published on October 5, 2022 by Andre Staltz
Last month went into fixing crashes. The biggest one is that database indexes shouldn't get bugged anymore (which was causing incorrect content to show up when you open a thread). This has been happening since March 2021, and it was a sneaky bug, very well hidden. There are other crashes we fixed in this release (0.2210.3), the full list being:
- ✅ Bug fix: broken database indexes
- ✅ Bug fix: crash when reindexing with live
- ✅ Bug fix: crash when seeing invalid post without text
- ✅ Bug fix: on Force reindex, go to Central screen
- 🎉 Feature: delete account button in the Settings
- 🎉 Feature: setting to opt-in for crash reports
- 🔷 Improve Chinese (Simplified), German, Persian, Russian translations
- 🔷 Deprecate "Enable detailed logs" setting
Manyverse has had too many crashes, across all 5 operating systems, but we were capturing crash reports only for Android thanks to the open source and local-first crash report library ACRA. For all the other operating systems, there is no equivalent to ACRA, unfortunately. On one hand, we thought about building such library, but it would be a large endeavour. On the other hand, we want to avoid cloud services as much as possible, but this decision was hurting us. Hurting users. So we decided to bring in the big guns. There are world-class SaaS products dedicated at solving exactly our problem.
We integrated Sentry.io's solutions in Manyverse version 0.2210.3, in such a way that respects user choice. On the app, Sentry's components are open source. In the cloud, they're proprietary. So this is why your Manyverse will only communicate with Sentry's servers if you enable it via the Settings screen. It's opt-in. Think of it as contributing to Manyverse. Money is not the only way you can help us make progress, crash report data is also a valuable resource. Also, Sentry was kind enough to offer us a free account on their cloud, so that's nice.
Manyverse is trying to achieve freedom from cloud and corporate servers, so a valid question is: what's the role of cloud services in Manyverse? And will this become a trend until Manyverse is centralized? The answer to that is that the core functionality in Manyverse should always remain – by default – free from the cloud. You can choose to enhance your app with servers, be them SSB room servers hosted on DigitalOcean or Sentry's report system, but that's your choice. And the enhancement must remain peripheral to the core experience of using Manyverse.
We have already relied of GitLab's cloud for hosting our source code, Figma for designs, GitHub for hosting our releases, Netlify for hosting our website, and OpenCollective for donations. These services allow us to be incredibly efficient. If we were to build and maintain all of these ourselves, I honestly don't think we would even have an app available for download. But we use these services knowing that we can move our source code away from GitLab if needed, we can move away from GitHub or Netlify or OpenCollective if needed, and so forth.
I have some happy news today, too! We sent out a survey to gauge user's satisfaction with the app, and the results indicate that people are overall experiencing improvements in the app. We got 18 answers. Before I show some statistics, here are a couple qualitative answers that bring us joy:
- "Manyverse has been solid for me. I'm happy with the change from Patchwork."
- "The UI has improved, as well as performance"
- "Indexing was too slow to get up-to-date, am now a regular user"
- "Thanks for all your incredible work! 🥰 Had a good time onboarding users to Scuttlebutt using Manyverse at DWeb Camp 2022. 🍑"
The distribution of installations across platforms was:
- 11 Android users
- 9 Linux users
- 4 MacOS users
- 3 iOS users
- 1 Windows user
When asked if we're making progress, people answered:
- "Much better" 2x
- "Better" 9x
- "Same" 3x
- "Worse" 0x
- "Much worse" 0x
To summarize the score above in a range from -2 to +2, we're improving at +0.92.
Next, we wanted to know if people noticed the new features we delivered, and whether they were useful features:
- 66% respondents noticed the "Friends in common" counter on the Profile
- 100% of those who noticed it also thought they understood what it meant
- The feature was "Useful" or "Totally useful" (score 1.5 in a range from -2 to +2)
And about the "Occupies" counter, people answered:
- 88% respondents noticed it
- 100% of those who noticed it also thought they understood it
- The feature was on average "Useful", but with mixed reactions (score 1.0 in a range from -2 to +2, with 1.3 as standard deviation)
I uploaded the raw survey results for all to analyse them, and I hope we can keep doing these surveys. I've already learned some interesting things, such as the Occupies counter not being that universally accepted as useful, and that the "Friends in common" counter is not easy to notice. We'll try to send a survey every time we release a feature, to get quantitative feedback on that feature.
We're making progress on redesigning the Public feed, and Andrew is already implementing some of the new features. Too early to share more about it, but this means you'll get a juicy new feature soon! For private groups, we're making progress in code and in specifications, and we believe that in November we can share our first milestone on the Manyverse blog. Stay tuned! And thank you for supporting us!