Control: tags -1 - moreinfo On Wed, Apr 01, 2015 at 12:44:33PM +0300, Dmitry Shachnev wrote: > On Thu, 26 Mar 2015 09:55:40 +0100, chrysn wrote: > > on a system on which qt4-qmake is installed as i386 package > > (4:4.8.6+git64-g5dc8b2) and qtchooser is installed as amd64 (no idea why > > this happened on my system, but qt4-qmake being Multi-Arch:foreign > > indicates it should be ok), qtchooser fails to find qmake: > > Does ???qmake -qt=qt4-i386-linux-gnu??? work for you?
Most likely, it does. The question here is: Does every consumer do that? Out of 1000 source packages failing to cross build, 256 fail due to something qtish. None of those fail executing QT stuff, most are due to not finding some required file. Suffice to say, even if -qt=qt4-i386-linux-gnu works, it is never passed in practise. As such, the expectation that people pass the flag is not realistic (now). We need to figure out what a reasonable expectation here is: a) If I install qt5-qmake:$somearch, running "qmake" should do something reasonable. This is not the case and the reason for this bug report. b) If someone installs qt5-qmake:$somearch, one must run it via "qmake -qt=qt5-$gnutriplet" to expect something reasonable. Clearly, nobody is doing this. So the first question here is: How does one correctly invoke qmake. Depending on the answer, we may have *lots* of bugs. I kinda prefer variant b, but like to confirm before filing 200+ bugs. ;) Whatever the answer is, please write it down in some debian/README.multiarch. Helmut
