ptitSeb
Serial Porter
I confirm compile QT 4.8.4 on the Pandora takes time. A day (or 2, I don't remember). But this kind of compilation time are just for a few monsters, for normal program, compiling on the Pandora is fast, and as you said, it's easier than cross-compile._wb_: good to know. I could work with that. I haven't really looked into the code::blocks PND (I assume that's what you mean?), but that sounds promising. I could try to compile some of my C++11 stuff with it to see if it fits the bill.
Getting Qt to compile (at least in the later 4.8.x versions, using ivanovic's toolchain because yactfeau had old gcc) for pandora has been a huge PITA (to get right), though I guess it would be somewhat simpler if using an actual environment instead of cross compiling. It would take a lot of time, possibly days, to compile though . Distcc would probably be an ideal solution here. With a good workflow (normal: dev on desktop, git pull to pandora, compile, test -or- bug hunt/release testing: ssh/scp to pandora, tweak, test, patch git + create PND + release) I'd probably be less hesitant to do stuff on it. That wouldn't mean fulfilling all hopes and wishes and spending all available dev time on it, but at least the small stuff and bug fixes.
Regarding Qt5 I actually think Panorama and PNDManager would benefit from it a lot. The new graphics system should be quite a bit more performant and actually threaded (less UI hangups). The more modular design should allow for a more fine-grained cherry-picking of Qt libraries resulting in smaller PND size -> shorter loading time. Moving Panorama and the UIs (PNDManager for one) to Qt5's QML2 and SceneGraph would be quite a bit of work, but it's work I'd actually be willing and interested to do . Optimally I'd like to see Qt5 in the firmware. There's a good reason some of the latest mobile platforms like sailfish and ubuntu touch have picked it.
QT5, I haven't look at that at all. QT4 & QT5 can cohabit?