Re: Various problems and/or questions on Insight 5.2.1

[Keith Seitz <keiths at redhat dot com>]

On Fri, 30 Aug 2002, LE GALLO Pierre OF/DSI/EXT wrote:

> I'm working on WinXP/cygwin, uname gives 1.3.12(0.54/3/2). Insight is 5.2.1

Eew. You're a glutton for pain, eh? ;-) (Cygwin + Tcl/Tk8.3 from 
sources.redhat.com = _PAIN_)

> 1) Insight build ok, and gracefully comes to life. BUT wish83 crashes. Saw a
> patch from Keith somewhere, but
> I suppose it's now in the current release, no? Wish83 from Khan works, do I
> have to go into modifications K. made?

Yes, I've committed those patches. wish should no longer die.

> 2) very strange behaviour of anchor (this one is special for Keith) in label
> creation in srcwin.itb: even when created
> without west orientation (center is default), label got left justified.

I'm sure that its a merge casualty.

> 3) I stole libgui and gdbtk library (big parts of them) to put in my own
> debugger for Akeso project. Well, I built itcl/itk libraries, but some bugs
> remain (menu inverse
> incorrectly and the like). Called from wish (Active or Khan patch) work
> correctly, so problem comes from my build.

Ugh. That means that more than just tcl and tk was messed up by the 
"merge". [You know, I've often considered starting over with the "port" to 
cygwin for 8.3 -- or with Mumit Khan's work. It can't be nearly as botched 
as what we have now.]

> When I try :
> gcc $(CFLAGS) dakeso.c -o dakeso -lcygitcl32 -lcygitk32 -lcygtk83 -lcygtcl83
> produce a very neat core dump! CFLAGS is -g and includes.

Make sure that you are starting up and initializing the cygwin DLL 
properly. See gdb/gdbtk/plugins/rhabout/rhabout.c for an approach
which seem to work. (BTW, if you're using cygwin, please be aware 
of licensing issues. I don't know if your project is personal or 
commercial, but I thought I'd mention it to be safe.)

> To resume, I'm looking for a Khan like implementation WITH itcl/itk, for it
> seems to me insight builds......only insight!

I would be very, very grateful if someone would compare what Mumit did 
with what we have checked in. Maybe someone will find out why our versions 
are so problematic.

(BTW, can you send me a pointer to Mumit's work?)
Keith