I have to strongly disagree with this (just like people who know what was behind the decision).

The upgrade to the next stable Qt version was nothing that Jolla wanted per se, it was a necessity.

First, they originally wanted to go with Qt4 and the whole UI was written in Qt4, but then the ST Ericsson disaster happened and they had to rethink their whole strategy and pretty much rewrite the whole OS, to take advantage of libhybris (as there were no manufacturers that would provide other than bionic drivers) - that required the switch from X to Wayland. Qt4 wasn't ready for Wayland, so they had to go with Qt5. At that time, Qt5 wasn't ready to ship on mobile devices, as many APIs were unstable.

So Jolla contributed a lot to Qt5 at that stage, but they played the fair game and instead of forking it (like Canonical sometimes does), they contributed to Qt5 upstream. But that meant that every time new version came out (with changes they helped to push), they had to backport these changes to their Qt version, which took more and more effort and became unsustainable.


Second, maybe you don't remember, but there was a huge (yet understandable) uproar in the developer community about many APIs not being available in Jolla Store. No location-enabled apps, no Bluetooth, etc. . It resulted in many apps being released through OpenRepos instead and an overall unfavorable situation for Jolla Store native application environment. It was bizarre, because releasing an Android app in Jolla Store allowed developers to take advantage of more APIs than if they released a native app.

This was because the Qt5 APIs were unfinished and unstable, so if they were allowed in the Store, it would mean apps could break if the API changed in the next release.

So they had to wait for the APIs to stabilize (and also contributed to them in upstream), which most of them did in Qt 5.2. So once Qt 5.2 came out, it was necessary to upgrade, so the stable APIs could come to the Jolla Store.

Yes, the process of upgrading Qt to version 5.2 was very painful (it introduced many new things, some of which were reflected in the memory management, as you mention), took lots of resources which could have been used in different ways, as you suggest, but it was absolutely essential for the platform.


I agree with you on some other points you make, but I wanted to point this out.


PS: I remember many people shouting (rightfully, yet again without seeing the whole picture) how SailfishOS is a joke, because it doesn't even allow location-enabled apps in its store. You apparently can't please everyone.