4 ms·
It’s interesting to me that the lack of this feature has been one of the key differentiating features of browsers on desktop OSes compared to mobile browsers, s
by dmitshur 5y ago
It’s interesting to me that the lack of this feature has been one of the key differentiating features of browsers on desktop OSes compared to mobile browsers, such as Safari on iPads (and iPhones). It’s hard to do serious work on a mobile device that may silently unload a tab with unfinished work just because it was placed in background, whereas desktop browsers somewhat guaranteed that wouldn’t happen.
In a way this change probably narrows the gap between browsers on desktop and mobile OSes. Hopefully this is a good thing, as I expect this implementation will be smart enough not to unload tabs with filled in forms and such.
I could be wrong in this interpretation.
- dredmorbius 5y agoThis. Mobile devices are so utterly unsuited to actual productive use for this very reason. (I'm quite familiar with this, have attempted to use them in this manner for years, and am typing this on a Chrome/Android session I utterly despise.) The problem extends to other tools -- Termux, useful as it is, also randomly gets garbage collected and killed. Best I can understand, Android actually employs some AI-based memory management whose training goal seems to have been "how to maximally annoy dredmorbius". Someone's earned their KPI.
- iggldiggl 5y agoIt's even worse now that multi-process is a thing even on mobile, because it means that tabs get unloaded not just when the browser feels like it (practical quality of actual implementations might vary, but at least in theory the browser should know which tabs have been recently used, which tabs might be playing audio, etc. and so make a somewhat reasonable attempt of prioritisation which tabs should get unloaded first), but also whenever the OS randomly decides that one of your child processes (containing a random set of tabs) should die. From following the bug tracker this definitively seems to be a problem with Firefox now – I wonder how Chrome gets around that problem, though? Do they suffer from the same problem, have they found some neat workaround, or are they cheating and being specially whitelisted by the OS?
- dredmorbius 5y agoYep, this.
- Vinnl 5y agoIt might still happen in the future, but what's described in this post doesn't seem like it's breaking that guarantee: it will only unload tabs that would otherwise have crashed the entire browser, so in cases where that guarantee wouldn't hold already.