From: "Douglas Kennedy" <doug...@comcast.net> > Point taken >
I take it this means we can now drop all the discussions about the joys of Win9x development and the responsibilities and obligations (or lack thereof) of open source developers, and let those who are interested in running msp430-gcc on Win9x work toegether towards solving the problems. To summarise the situation, as far as I understand it (correct me if I'm wrong - it's always possible that I'm mixing things up with the avr-gcc project, which has similar problems): 1) There are no serious problems either building or running msp430-gcc and gdb (inluding proxy) on Windows NT or W2K (anyone tried XP? ). 2) Some people with Win9x have had things working, others have had problems. 3) Some builds work better than others on Win9x. 4) Cygwin is known to have "issues" with Win9x, but has been improving. 5) Mingw-compiled gcc works fine under Win9x, but mingw-compiled gdb (and, in particular, insight) have big problems with paths. It seems to me that there are currently several ways to build Windows binaries. Building the gcc toolchain and gdb under Cygwin will produce binaries that work fine on NT-based windows. Building the gcc toolchain under mingw will produce gcc tools that work fine under all versions of windows, but which have trouble with gdb and/or insight, mainly because of path issues (mingw tries to be windows-like with \ paths, cygwin tries to be unix-like with / paths). It also seems that the 2003-02-10 version of the windows install package works better on Win9x than the 2003-03-06 package. Can anyone say categorically which tools were used to build these versions? (For some reason, I can't connect to mspgcc.sourceforge.net just now). If the problem can be narrowed down to a matter of the tools used to build msp430-gcc, then it should be possible to post different builds at sourceforge. Perhaps the Win9x version will have reduced functionality (such as lack of insight - maybe gvd will work instead?), but it would be better than nothing.
