4 ms·
Crashlytics is necessary unless you want shite software.
by iamstupidsimple 6y ago
Crashlytics is necessary unless you want shite software.
- panpanna 6y agoRight now we have shitty software _and_ a crashlytics that phones home constantly for no apparent reason.
- yjftsjthsd-h 6y agoThere's a rather large selection of good-to-extraordinary software that doesn't use any sort of analytics/auto-reporting. Off the top of my head, GNU coreutils, and BSD flavor, sqlite, qmail, (neo)vim, emacs, GIMP...
- iamstupidsimple 6y agoThose are all CLI/desktop apps and stable as they've been around for decades. They also had supportive communities to file bug reports, changelists, etc.. You can hardly compare coreutils to smartphone apps. As someone who deals with Android fragmentation daily, you wouldn't believe how hard it is to support the whole ecosystem. Disliking analytics I can understand, but crash reporting adds measurable improvement in the quality of software.
- yjftsjthsd-h 6y agoYeah, I realized halfway through my list that I was mostly listing infrastructure; that's why I tacked on editors and GIMP. The original claim was that analytics were required to avoid software being terrible, and I mostly wanted to point out that this is obviously false; kind of software and community support make an impact, but they're not a hard requirement, and at any rate I think we're moving the goalposts here. > You can hardly compare coreutils to smartphone apps. Indeed; coreutils works on an absurd variety of unix-family systems, is copyleft, respects privacy, and is high-quality. If smartphone apps were more like coreutils, we'd all be better off. > As someone who deals with Android fragmentation daily, you wouldn't believe how hard it is to support the whole ecosystem. Disliking analytics I can understand, but crash reporting adds measurable improvement in the quality of software. I do understand where you're coming from; while I don't have skin in that game, I'm passingly aware of the remarkable ways that vendors break their systems. And honestly, if you get user consent, I don't actually have a problem with having the app report problems to you. It's just the "send information about the user's system without asking" thing that I object to.
- viraptor 6y agoThey don't need to auto-report because the system takes care of that: https://wiki.ubuntu.com/CrashReporting https://wiki.ubuntu.com/CrashReporting https://developer.fedoraproject.org/tools/abrt/about.html https://developer.fedoraproject.org/tools/abrt/about.html and others. Gimp: https://bugzilla.redhat.com/show_bug.cgi?id=1828800 https://bugzilla.redhat.com/show_bug.cgi?id=1828800 Emacs https://bugzilla.redhat.com/show_bug.cgi?id=1763503 https://bugzilla.redhat.com/show_bug.cgi?id=1763503 Neovim https://bugzilla.redhat.com/show_bug.cgi?id=1826357 https://bugzilla.redhat.com/show_bug.cgi?id=1826357
- ta17711771 6y agoIt's almost like you can ask your users to help with automated one-touch bug report submissions, instead of taking the data without asking.
- yjftsjthsd-h 6y ago> Apport will notify you about the crash and offer to file it. Do that. I wouldn't call it auto-reporting when the user has to agree first; consent is the sticking point. (Same for abrt) (granted, I should have been more precise and I suppose that is automated in a sense; I took "automatic" to mean "without asking")