[webkit-qt] QtWebKit --minimal build with minimal Qt

[noam.rosenthal at nokia.com]

On Sep 27, 2011, at 3:27 PM, ext Andreas Kling wrote:

> Heyo, interesting topic!
> 
> In my opinion all QT_NO_* defines should be dropped from WebKit. I believe QtWebKit is sufficiently complex that you shouldn't use it without a fully-featured Qt.
> 

It only adds to the complexity matrix if we have to support them. But if we allow them in without actively supporting them, letting the community fix problems if they arise, I don't see the downside of keeping them.
We've actually been doing that for a while now; some brave souls like Suzuki san were fixing issues with QT_NO_* when they arise; that doesn't add any commitment or cost to anybody else. Especially if all we test in the bot is that it builds, not even run tests.
Allowing a low-footprint webkit scratches the itch of many people doing great things; Let's let them keep doing that, while being smart about the cost (which is IMO negligible in this case).

> For every additional build-time flag, the complexity matrix grows larger, and we're not testing any of it beyond building with all those flags enabled together. If we do need to support that, I'd rather we had a monolithic USE(LESS_QTWEBKIT) flag rather than separate ones.
Those end up being kitchen-sinks… I prefer the current situation. Can you refer to one instance where the QT_NO_* stuff created issues for someone rather than the people who care about it?