Keith Seitz wrote: > I noticed that you're having to distinguish between windows and mingw > in a lot of places... I think I'm going to have to whack the use of > "tcl_platform(platform)" in favor of a procedure that returns > "cygwin", "mingw", or "unix". If I check something in quickly, would > you be adverse to updating your patch? Or I could wait until after > this is all committed. > I'm not sure that would help - I had considered just stubbing ide_cygwin_path on mingw so there was no path conversion but back in this message - http://sourceware.org/ml/insight/2007-q2/msg00112.html Chris Faylor said that cygwin uses the /dev/comN format for serial ports. With a windows native binary produced by mingw we can only access the COM form. Checking for the existence of the ide_cyqwin_path command gives us a handy way to differentiate when setting up the port list. I'm not sure but I think splitting "windows" into "mingw" & "cygwin" would complicate things a little more, unless I misunderstood what you're suggesting. > I don't see any problems with your changes to the insight code base. > Your changes there (although numerous) are small/trivial enough that > I'm going to forgo asking you to jump through the hoops of getting an > assignment for Red Hat. > > I'll also approve the changes to tcl/, but I'll have to take your word > for it, since I don't have any way to test this. Hopefully a more > cygwin-savvy developer out there will chime in with any comments. > > I'm curious: have any of these Tcl mingw patches made their way > upstream? Do they even care any more? The latest Tcl/Tk sources build OOB for mingw, I just took enough to fix the problem - the hint originally came from Andrew Stubbs when he posted the patch back at http://sourceware.org/ml/insight/2005-q3/msg00056.html The relevant file is http://tcl.cvs.sourceforge.net/tcl/tcl/win/tclWin32Dll.c?revision=1.49&view=markup (SF changed their cvs host naming since Andrew posted that message) Most of the Tcl patches are in the gdbtk library for Insight relating to path conversion. Hopefully the attached patch & changelog is suitable, thanks again for all the advice. Dave