Hi Dan!
Great that you are working on a Windows version of the Analysis software.
It surpriese me a bit that you have gone for X-windows emulatoin under
Windows. I do not know that much about tcl/tk, but I thought the idea
behind it was that you should be able to run programs under different
"windowing" systems?
What is it in Analysis that is X-dependent? There is a bunch of other
software out there running under multiple "windowing" systems. One of
the best ports I have seen is MOLMOL, but even Sparky (which I believe
uses Python and Tcl/tk) runs under Windows, and do look and feel the
same as under Linux.
I have been working a bit with OpenGL under Windows (but not under X)
and I thought the main differenve is windows handling, updating and
"double buffering"? Winodws also has a few extra features for OpenGL,
you can for example display text in OpenGL rendering contexts.
What about a tcl/tk version of analysis (without OpenGL), would that
be possible?
If you need more input, do not hesitate to post to the list!
Best regards,
Magnus Helgstrand
2005/11/10, Dan O'Donovan <[log in to unmask]>:
> > has anyone ever attempted to compile Analysis for Win32 PC systems?
>
> This is something I've been working on on and off for the last couple of
> months. The formatConverter certainly works under Win32 and has been
> working for some time. Unfortunately the Tk/Tcl and OpenGL combinations in
> analysis throw up a few serious bugs when compiled in Win32.
>
> I'll quickly mention a few of the problems to see if anyone has any
> ideas... To compile the C code I use either cygwin or MinGW (with the
> appropriate --mno-cygwin flags etc.) I use these compilers as some of the
> Tk/Tcl parts of analysis rely heavily on the X window system - which isn't
> part of the native Win32 system, but is included with these compilers (and
> not Visual Studio).
>
> When running the Tk parts of the code the program falls over due to what I
> believe is a bug in the cygwin X implementation. It's something to do with
> a function called XCopyArea() that doesn't do what it says on the box. I'm
> working on getting some sample code to replicate the error so I can
> submit it if it really is a genuine bug.
>
> To get the OpenGL code to work (I believe) would take a drastic overhaul
> of the OpenGL code and result in incomapibilities with the original ccpn
> distribution. (The Win32 implementation of OpenGL seems to be different
> enough to cause headaches for people working in both environments).
>
> If you'd like me to share my Win32 Makefiles, include files and scripts
> let me know. I could also try and walk anyone through the slightly painful
> compilation if they fancy it.
>
> If you really really want to run analysis on Win32 *today* then you might
> like to look at some of the X-emulators available for Windows. However I
> don't think this is something anyone has tried before and would probably
> result in a severe performance hit!
> Using something like putty it is possible to ssh into linux boxen and
> bring up their X programs on your Win32 screen. However, this can get a
> little confusing when you have to transfer your projects from machine to
> machine...
>
> --
> --------------------------------------------------
> Dan O'Donovan CCPN PhD student
>
> 'Of course all bell-ringers are eccentric. Do you
> think we'd do this if we were normal?' JRF
>
|