From ruby-talk Mon Mar 12 17:18:44 2007 From: "Uma Geller" Date: Mon, 12 Mar 2007 17:18:44 +0000 To: ruby-talk Subject: Re: QTRuby + QT4 + Win32 Message-Id: <6f9697050703121018p26f6c91fx227ebaa53435ff5e () mail ! gmail ! com> X-MARC-Message: https://marc.info/?l=ruby-talk&m=117390716129388 I've been using Linux openSUSE 10.2 and it already comes with Qtruby prepackaged, and great support for Ruby development in KDevelop, and I really love it. Richard's work is impressive, kudos for Richard. The problem is, I also have customers who live in Windows and honestly, there's certain tasks where a web interface just won't cut it, no matter how polished your AJAX or Flash patch-tech is. They will move to Linux or some Unix eventually, but the burden is on us the developers to build the necessary bridges for them to cross the chasm. I have a relatively acceptable GNU tools system installed and working on Windows, crafted on top of Msys and other packages I gathered from around the web. The problem is there is not a single regular incompatibility pattern you see repeating everywhere: that would take effort, but would be a solvable problem. The problem with ported apps on Windows is that different applications and libs build things differently under different assumptions. For example, most GTK apps rely on pkg-config to tell whether certain packages are installed. Many cross-ported apps from the unix/linux world sometimes assume Cygwin is the only posix-like system on Win32-land, some others (like QT4) rely on a Mingw32-only system. This leads to an neverending There's just no way a one woman effort can patch the entire codebase of ports to make it work on Windows, even too complex to document it. This is not a Windows problem, this is a problem about assumptions and lack of know-how by porters or unix/linux developers targeting the Windows-based platforms (for a moment I will be forgiving and will omit other possible reasons I have considered) The build systems they choose for a given package aren't interoperable. This problem is calling for a resolution, and it's probably even beyond GNU's abilities to address it, it takes a concerted effort from all developers targeting one of the Win32-based platforms, be it mswin32, cygwin, or mingw32. Richard probably knows a lot more about build systems than I do, and he's expecting CMake to address this pain, so I'm confident we'll really have it solved in the near future. That being said, if CMake is good enough to solve the problem, all porters and unix developers should review their build systems. So we really are talking about two different problems of different size and scope here. 1) Building QTRuby for Windows 2) Fixing the build system so that Unix/Linux developers have a single target to port for. On 12/03/07, Caleb Tennis wrote: > > > > Has anyone managed to compile QTRuby QT4 > > for Windows ? (Open Source edition) > > If you can get the build system installed on Windows, I can probably > help with the rest of the steps. I tried installing the build system > via the steps here (at least, the first 8 or so): > > http://kdelibs.com/wiki/index.php/ > Building_KDElibs_4_using_GCC_and_MinGW_with_Qt_4.2 > > But installing the GnuWin32 didn't work for me and I gave up. > > > Maybe I'll add some instructions for how I build qtruby qt4 to the > KDE techbase. I think that will apply here as well > > Caleb > > -- best, UG --- Uma Geller http://umageller.wordpress.com