[GRASS5] Re: [GRASSLIST:3815] openOSX again

On Jul 9, 2004, at 12:33 AM, grass5-request@grass.itc.it wrote:

If these problems continue, perhaps we should remove mention of OpenOSX
from the GRASS website and more heavily promote Lorenzo Moretti's great
work[*] as the situation as it exists reflects badly on the GRASS
community. Their link to the GRASS website leads to a non-existant page
Baylor Univ.

see: http://grass.ibiblio.org/grass50/binary/mac_os_x/

These issues are somewhat disturbing. I agree that they could reflect poorly on GRASS. It is too bad, as OpenOSX potentially is a very nice alternative way to get open source applications onto a Mac.

[*] http://wwwamb.bologna.enea.it/forgrass/

Perhaps we should do that regardless of any problems people have with
any 3rd party vendor.

IIRC, updating things so people can use the Mac/Aqua version of tcltk
for the tcltkgrass menu system was only minor changes?

see http://article.gmane.org/gmane.comp.gis.grass.devel/2681

& in response to the problems listed there:
1. The web browser bit is already done for 5.7's gui help system.

Also GRASS uses the currently selected web browser (at least in OSX 10.3) in 5.3 if you enter the term 'open' in the browser name of the g.manual dialog. This doesn't seem to work for the ctrl-middle button style help for each command, but perhaps there is a way to configure this in the 'configure html' dialog.

2. The xwininfo problem is solved in 5.7 by d.info.

This does not seem to be an issue for current versions of tcltkgrass. The major problem that I recall with tcltkgrass under 5.3 (and maybe 5.7) is that the menu bar wants to incrementally increase its vertical dimension each time you start tcltkgrass. I *think* this is tied to the use of wm geometry somehow, but could never figure it out. I 'fixed' it with a hack that does a menu resize each time you close tcltkgrass.

3. easy to change

Where is this an issue?

4. no action needed??

This does not seem to be an issue.

maybe Michael Barton has already fixed some of these issues
incidentally?

Prior to Lorenzo's packages, the main issue from a user interface/GUI perspective was finding a simple way of getting the libraries needed to run GRASS and tcltkgrass. The only options were compiling from source, using Fink to compile (but sometimes the libraries were out of date), or using OpenOSX. Marcus provided a set of binary Proj4 and GDAL libraries, but one still had to be able to go into the terminal and use the sudo command to install them. TclTk remained problematic.

By making these available in a single package, Lorenzo solved this issue. Furthermore, by putting all the binaries into Mac pkg format, Lorenzo has made GRASS as easy to install as OpenOSX. More importantly, he has kept up with GRASS changes by making versions 5.3 and 5.7 available (when I'm back in the office at the end of the month, I'll be able to update my mirror), and incorporating updates to tcltkgrass.

AFAICT, GRASS runs as well under OSX now as any other platform and better than most, judging from the discussion on this list. The main difficulties I've run into using/building a GUI under 5.7, for example, are ones that are problematic across all GRASS platforms (the G_PARSER() problem, for example). For Mac users, I can't see any advantages in using the OpenOSX package now, and several disadvantages.

NVIZ is another story...

My understanding is that NVIZ would need some significant changes to be able to work with a non-X11 version of TclTk. Given this and the current uncertainty over the direction of the next generation GRASS GUI (I've heard several interesting ideas proposed), it does not seem worth the effort to invest in a special Mac version that uses a native Aqua/TclTk combination. I'm all for doing an Aqua version (though using X11 is a non-issue as Apple provides it in pkg format for easy installation); it is the TclTk part that I'm uncertain of. If GRASS ends up sticking with TclTk, I am happy to work in that direction. FWIW, I think that cross-platform portability is a strong point of GRASS and should be kept as much as possible.

fwiw, we purchased the OpenOSX CD in 2002 or so when the book first came
out and had no problems with them back then.

One of my Spanish colleagues bought a CD last fall (2003). It installed very well. However, in running it, I ran into several annoying bugs and lockups. I don't know whether these were a function of using 5.0.2 (instead of 5.0.3) or something in the OpenOSX setup. We switched to a Fink version of TclTk and a current GRASS binary (5.0.3, then 5.3), and everything ran smoothly.

Michael
____________________
C. Michael Barton, Professor
School of Human Origins, Cultures, & Societies
PO Box 872402
Arizona State University
Tempe, AZ 85287-2402
USA

Phone: 480-965-6262
Fax: 480-965-7671
www: <www.public.asu.edu/~cmbarton>

Michael Barton wrote:

> NVIZ is another story...

My understanding is that NVIZ would need some significant changes to be
able to work with a non-X11 version of TclTk.

Someone would need to write an OSX (agl*) version of the code in of
Togl.c. It already includes both X11 (glX*) and Win32 (wgl*) versions,
so it should be reasonably clear which parts are portable and which
are platform-specific.

The underlying OGSF library shouldn't need any changes.

--
Glynn Clements <glynn.clements@virgin.net>