Windows binary saga

[Dirk Hohndel]

Just to keep people updated. After a complete A-Z rebuild of every library
and component from source (I literally removed the build directory
including the MXE installation and started from scratch) I am nowhere.

Things compile, installer builds and runs successfully on a Windows VM.
(this requires some manual intervention as otherwise QtWebKit isn't
built).
But subsurface.exe then crashes pointing at QtCore as the culprit.

Both Thiago and I tried to figure out what's going wrong there and were
not successful.

Based on another suggestion from Thiago I am now trying to compile a
debugging build of Qt5.5 for Windows but that keeps failing. It is
explicityl turned off for MXE - the reason given should not cause us a
problem, but my guess is there are other problems hidden by the fact that
these builds are normally turned off and therefore not tested with MXE.

I thought about building natively on Windows... I actually went out and
bought a Windows 10 machine today in part to debug things (assuming that
I'll have more luck there than with just a VM) and in part to re-evaluate
native builds. But of course that breaks all of my build automation and
most everything else that I have worked on over the past few years for our
Windows builds.

So I guess the real solution is to continue to track down
a) why I can't create debug builds of Qt5.5 under MXE and
b) why the release builds that I can create then crash when subsurface.exe
is run under Windows.

To make matters more interesting I am leaving on a trip on Friday.

So at this point I guess I need to declare 4.5 on hold.

Any help would be extremely welcome. I'll try to check in my latest build
script with steps to recreate the complete build environment.

/D