Re: Lin32Lib?
- Posted by David Cuny <dcuny at LANSET.COM> Apr 21, 2001
- 450 views
jbrown wrote: > So? Why not have Win32Lib fake a canvas? And break all those existing programs? It seemed better to start from scratch with a new library. But writing a multi-platform library is a never-ending process, and I ultimately decided that finding a pre-existing cross-platform library (wxWindows) was a better plan. But that's another long story... There's another twist to the story: almost all the GTK development was done in Windows. Someone did a port of the Gimp to Windows, and you can use the DLLs to write GTK apps. For me, it was a lot easier than working in Linux, since I'm more familiar with the DOS prompt than the Linux command line. But this caused me a bit of grief. I had written a Euphoria program to read the GTK include files, and generate wrappers for the structures. But apparently because of alignment issues, the position of structure fields is different under Windows than it is in Linux. That meant that some of the code would fail - seemingly randomly - under Linux. > I though that's why you were working on Llama, > which wouldn't have all that extra Win32 specific > stuff, but still support the general Win32Lib > framework. Yeah, that was the plan. I didn't take into account how much of a pain it would be to do. Ultimately, I decided it wasn't really the optimal way to go. > P.S. When I tried to work on Llama/GTK, most of > my changes were ineffective. Odd. > For example, I couldn't figure out how to pass > a char array to gtk_init, or why gtk_entry_set_text > keeped saying it's parameter wasn't a GTK_ENTRY type. Well, GTK has it's own type system built in. > Is this because of Euphoria's inability to > use structures and such, No, there are lots of programs (like Win32Lib) that interface C strutures. > GTK's assumtion of interfacing with C code No, also not an issue. There are lots of counterexamples to that, too. > or because of an improperly wrapped GTK? Most likely. -- David Cuny