Qt Webkit - Windows CE - Qt 5.3.x
-
I'm trying to build Qt 5.3 (including Webkit) against an Arm v4 architecture for Windows CE 6 using Microsoft's tool chain.
In short -- it is hard work.. Fighting against a weak and buggy C-runtime.. Fighting against an outdated OS. Fighting against an Windows CE OEM who has left out essential functionality..
Anyway, I'm now trying to hoodwink ICU into building for the Arm target. Originally I tried manipulating the autoconf scripts to target Microsoft's Arm v4 compiler, but gave it up when autotools tried to execute a sample program. So now I'm tackling the problem from a different angle.
I ran ICU's configure script for the regular Microsoft x86 2005 compiler and am now manipulating the Makefile with some success.
Has anyone gone down this path? Yes, I'm stuck with the Microsoft tool chain (VS pro 2008)
-
I managed to build Qt5 Webkit for a Windows CE ARM target using Microsoft's tool chain. It was a very messy build, which required source code modifications to Qt, ICU, and Webkit.
The modifications were generally superficial, but it's clear that Windows CE and Microsoft's tool chain are barely adequate for the needs of modern libraries. My driver was an industrial product where lifetimes are measured in decades rather than years, so the effort was worth it
Loosely put:
- build Qt5 core using the regular method
- build ICU via cygwin by first configuring for MVSC + NT then modifying the makefiles and source to suit the MSVC + CE tool chain
- build Webkit using the regular method
I'm in the process of testing and cleaning up my very ugly build scripts, but I don't think it is something that could be pulled into Qt or ICU or Webkit. It's just too messy.
What's the best way to share this with the community? ideas.
-
Hi and welcome to devnet,
You could e.g. create a wiki entry about that. But you could also ask on the development mailing list whether the patch may be of interest.
-
I have some results from field testing on a low powered device (500MHz, 256MB RAM)
Initialization is way too slow. And I seem to have introduced some serious regressions in the applications web UI, which may well be due to the code I had to rip-out / replace in Webkit and ICU.
Anyway, I've been asked to pursue an alternative approach based on adding Freetype support for Windows to Qt4. Field testing is promising (no observed regressions). See "this code review":https://codereview.qt-project.org/#/c/100430/ for more
-
I have some results from field testing on a low powered device (500MHz, 256MB RAM)
Initialization is way too slow. And I seem to have introduced some serious regressions in the applications web UI, which may well be due to the code I had to rip-out / replace in Webkit and ICU.
Anyway, I've been asked to pursue an alternative approach based on adding Freetype support for Windows to Qt4. Field testing is promising (no observed regressions). See "this code review":https://codereview.qt-project.org/#/c/100430/ for more
