On Wed, Jul 26, 2017 at 11:05:57AM +0200, Emilio Pozuelo Monfort wrote: > OK I see that porting is already done upstream in the link you quoted. So > yeah, it'd be a matter of moving to that. This probably needs a library > transition though.
The problem is that wx's GTK3 support is still much buggier than its GTK2 support. Upstream wx seem quick to fix bug reports about GTK3 shortcomings when they can reproduce them, but they've been unable to reproduce some which I can reproduce in their own sample code, e.g.: https://trac.wxwidgets.org/ticket/17915 So this transition is going to be a lot more work than tweaking wxwidgets3.0's debian/control and debian/rules, uploading and then bin-nmuing 100 or so packages. It's going to need a lot of testing of rdeps, and that's likely to uncover a significant number of upstream bugs which we'll ideally need to find reproducers for. My plan is to put a build of wxwidgets3.0 using GTK3 in experimental (or maybe unstable) so maintainers of rdeps can try it and we can gauge how realistic switching is in this release cycle. But as I've said already, I don't have a problem with just disabling webview before buster if it comes to it. It just seems unhelpful to boinc users running testing to do that right now, unless you're actually at the point of removing the old webkitgtk from testing. Cheers, Olly
