[winGRASS] cygwin and winGRASS status update

[Mike Thomas]

Hi all.

> But i have to dissapoint you that the generic GRASS windows driver would
> not solve all the problems installing and running GRASS on cygwin. I
> looked what it needs to create the driver and found that i do not know
> enough about Windows GUI and cygwin sockets programming to do that. And
> i fear that it may be a waste of time, because of the uncertain future
> of the GRASS X Driver and Windows beeing sort of a moving target.

Yes, this is a problem.

>
> I think that it is very important to maintain a shared code base for all
> GRASS "ports", including the cygwin port. Otherwise we would get a
> complexity we are not able to handle and fixes to the GRASS CVS would
> not be integrated into the cygwin port.
>

One way around this problem is to use a cross platform GUI library which
does not depend on X WIndows under Win32.

Over the past few days I made a very simple (dumb and unfortunately
not-yet-working (socket not available)) clone of the PNGDriver using the
Cygwin port of GTK, but then I realised it is dependent on X anyway.  There
is also a W32 native port of GTK, which might be usable via DLL from
Cygwin - haven't tried this one yet - probably can't for another week or two
now.

Other GUI possibilities include TK, GLUT and WXwindows.  The latter is very
successfully used by Rice University for PLT on multiple platforms.

GTK has the advantage of being modelled closely on X WIndows, so it might
make it easier to port the XDriver.  I don't know the other options well
enough to comment on this point in relation to them however.

Whichever - the end result is unified sources without having to muck around
with XWindows on Win32.


> The GRASS source and tcltkgrass is extremly interwoven with UNIX style
> utilities, so that porting only the X Driver would not help using
> tcltkgrass on cygwin. Remember that tcltkgrass uses xwininfo, xterm and
> other UNIX X Window utilities, which have no equivalent in Windows.

Maybe someone has to bite the bullet and embed all the functions in libgis
into native builds of TCL.  This might also give another way of making a
cross platform display via TCL/TK.  A lot of work though.

>
> And i have to stress that i had a perfectly running GRASS on cygwin with
> the Starnet Demo Server until the update for cygwin broke the X Driver.

Yes, this was very annoying - they seem to be slow on updates - must be
focused on GCC 3.0.

> As i expect that the Cygwin/XFree86 system will improve further in the
> next time, so i want to propose that we first try to isolate the problem
> with the X Driver on cywin.

> I don't think that a presale of the cd would be a good idea, as i am not
> sure if / when the complete cygwin port will be ready. And i think that
> we should spend money first on things that generally improve GRASS (map
> datum implementation etc.) than on things that only help windows
> GRASS users.

Especially the datum shift problem - this is pretty fundamental in a GIS.

Thanks for the effort Andreas and good luck with your move Markus N.

Cheers

Mike Thomas.