I read through the discussion yesterday of which version(s) of GRASS 5 should be released next. Given Helena's comments I decided to be bold (for me) and try to compile GRASS 5.3 from the current CVS snapshot. I'm running Mac OSX 10.2.8.
After puzzling over what turned out to be a syntax problem in my configuration parameters, I succeeded.
I've run through a variety of trials using the Spearfish sample data and everything seems to be working fine.
If it would help speed up the release to make it easier for more people to test 5.3, I am happy to make the OSX binaries available for distribution. I can explain how I configured the compilation, and could recompile it if something I did could cause problems for others. I'd probably need a bit of help in how to package up the binaries properly for distribution. Also, I haven't the foggiest idea of how to make the shell script that always seems to accompany binary distributions (though it basically seems to be an ungzip+untar command accompanied by a variety of measures to check if a system has the proper resources/directories to install and run grass). However, if its simply a matter of editing the 5.0.2 script a bit, I do it with guidance.
IMHO: As an aside and in response to the concern over the GRASS user base and attracting programmers to the team, there may be considerable 'pent-up-demand' for GRASS on Mac OSX platforms. After making GIS packages available for Mac users in the 1990's, both ESRI and MapInfo abandoned the Mac platform several years later. GRASS has a golden opportunity to expand its user base here. I am certainly encouraging this with the Mac users I interact with. Apple lists GRASS on its OSX web site. However, while learning GRASS is no harder (or easier) than ArcView or MapInfo, installing it is by a long shot. While I think Macs have a considerably larger installed base than any other particular Unix flavor, most Mac users are not Unix savvy. Installing from the terminal via a shell script is an adventure fraught with trepidation for most of them; compiling from source would not even be imagined by most. (The same is true of most MS Windows users of course.) Ideally, GRASS binaries for OSX should be released in package format (I probably have the tools to do this, but have not yet used them if I do) like the version that Open OSX has produced, and the versions of PostgreSQL and MySQL by Marc Liyanage. X11 should probably accompany it as an option, though it this will be less necessary in the future as Apple now packages X11 with its OS. Just some thoughts...
Un saludo cordial
Michael Barton
____________________
C. Michael Barton, Professor
Department of Anthropology
PO Box 872402
Arizona State University
Tempe, AZ 85287-2402
USA
On Sun, Nov 09, 2003 at 06:09:11PM +0100, Michael Barton wrote:
IMHO: As an aside and in response to the concern over the GRASS user
base and attracting programmers to the team, there may be considerable
'pent-up-demand' for GRASS on Mac OSX platforms. After making GIS
packages available for Mac users in the 1990's, both ESRI and MapInfo
abandoned the Mac platform several years later. GRASS has a golden
opportunity to expand its user base here.
The same is almost true for Unix based, since some major vendors have
simply dropped all support (for example Intergraph, which has made
some desastrous strategic choices in the past (leaving RISC and Unix
for i386 and being swept out of the market)... and now sticks to M$
products... It seems I'm not ready to be a CEO because my brains are
unable to understand some things...;).
If it would help speed up the release to make it easier for more people
to test 5.3, I am happy to make the OSX binaries available for
distribution. I can explain how I configured the compilation, and could
recompile it if something I did could cause problems for others. I'd
probably need a bit of help in how to package up the binaries properly
for distribution. Also, I haven't the foggiest idea of how to make the
shell script that always seems to accompany binary distributions
...
I'm interested in helping too. I made a binary distribution once before,
and using instructions from Markus, it created that script
"automatically", so that shouldn't be an issue. I'll forward you an email
I have from those days in a minute. In short, there were two ways: (1)
follow the directions in the documentation/release_rules.txt under the
main source directory, and (2) Markus has a script that he uses to get
through all the steps in the release_rules... but unless there are new
developments, this script is not completely portable. On the other hand,
it didn't take TOO much modification to make it work for me on my system,
and it helped make sure I was making binaries the same way as Markus.
The main "problem" with the approach is that it used statically linked
libraries which meant that the binary distribution was HUGE. Markus
recently succeeded to get the CVS version of GRASS to compile with shared
libraries, if I recall correctly. However, I don't have the information
on how he managed to do this. So, much as I hesitate to bother Markus
with more requests, Markus, if you have more up to date information and/or
scripts that you could share with us from your OS X adventures, can you
pass them on so hopefully we can help keep OS X versions built without as much
intervention from you ?
The main "problem" with the approach is that it used statically linked
libraries which meant that the binary distribution was HUGE. Markus
recently succeeded to get the CVS version of GRASS to compile with shared
libraries, if I recall correctly. However, I don't have the information
on how he managed to do this. So, much as I hesitate to bother Markus
with more requests, Markus, if you have more up to date information and/or
scripts that you could share with us from your OS X adventures, can you
pass them on so hopefully we can help keep OS X versions built without as much
intervention from you ?
GRASS 5.0/5.3 already supports shared libraries on Linux using the
alternate build mechanism in the "mk" directory; see mk/README for
more information.
You might want to see what would be involved in porting
mk/mid.mk.shlib to MacOSX. If the difference between creating a static
library and a shared library is simply a matter of:
a) compilation flags, and
b) the command to create a library from a list of object files,
On Sun, Nov 09, 2003 at 06:09:11PM +0100, Michael Barton wrote:
I read through the discussion yesterday of which version(s) of GRASS 5
should be released next. Given Helena's comments I decided to be bold
(for me) and try to compile GRASS 5.3 from the current CVS snapshot.
I'm running Mac OSX 10.2.8.
Very nice.
We can use more reports like this.
After puzzling over what turned out to be a syntax problem in my
configuration parameters, I succeeded.
I've run through a variety of trials using the Spearfish sample data
and everything seems to be working fine.
Good.
If it would help speed up the release to make it easier for more people
to test 5.3, I am happy to make the OSX binaries available for
distribution.
Please do so.
As long as they are clearly labeled
binaries will give some more people a chance to try stuff.
On Sun, 9 Nov 2003 13:27:00 -0500 (EST)
Scott W Mitchell <smitch@mac.com> wrote:
I'm interested in helping too. I made a binary distribution once > before, and using instructions from Markus, it created that script
"automatically", so that shouldn't be an issue. I'll forward you an > email I have from those days in a minute. In short, there were two > ways: (1) follow the directions in the > documentation/release_rules.txt under the main source directory, and > (2) Markus has a script that he uses to get through all the steps in > the release_rules... but unless there are new developments, this > script is not completely portable. On the other hand, it didn't take > TOO much modification to make it work for me on my system, and it > helped make sure I was making binaries the same way as Markus.
Perhaps you can use fink (http://fink.sourceforge.net) to provide complete grass binary packages.
Fink is like the debian package manager (apt-get) and there for very powerfull, as it can check dependencies and download the needed packages too.
Fink already provides 2500 open source packages for MacOSX.
A user would just have to do a
apt-get install grass
And that would download all dependent packates and install them for him.
Yes, I've been wondering about that too - I use fink heavily. There is already a fink package for grass 5.0.2, perhaps a 5.3 version would be a good idea too.
I'm all for choice, though, so a separate binary sounds good too.
Cheers,
Scott
On Monday, Nov 10, 2003, at 09:00 Canada/Eastern, Jens Oberender wrote:
Perhaps you can use fink (http://fink.sourceforge.net) to provide complete grass binary packages.
Fink is like the debian package manager (apt-get) and there for very powerfull,
...
------
Scott W. Mitchell Scott_Mitchell@carleton.ca
Department of Geography and Environmental Studies
Carleton University, B349 Loeb Building
Ottawa, ON Canada
+1-613-520-2600 ext 2695
But I am not updating it regularly.
Drawback: due to circular dependencies in XDRIVER I had to compile
it statically. So it's large. Unless someone fixes the XDRIVER
(see 'grass5' archive for error message) it won't be usage with
dynamic libs on MacOSX.
Cheers
Markus
On Mon, Nov 10, 2003 at 03:00:01PM +0100, Jens Oberender wrote:
Hi
On Sun, 9 Nov 2003 13:27:00 -0500 (EST)
Scott W Mitchell <smitch@mac.com> wrote:
> I'm interested in helping too. I made a binary distribution once > before, and using instructions from Markus, it created that script
> "automatically", so that shouldn't be an issue. I'll forward you an > email I have from those days in a minute. In short, there were two > ways: (1) follow the directions in the > documentation/release_rules.txt under the main source directory, and > (2) Markus has a script that he uses to get through all the steps in > the release_rules... but unless there are new developments, this > script is not completely portable. On the other hand, it didn't take > TOO much modification to make it work for me on my system, and it > helped make sure I was making binaries the same way as Markus.
Perhaps you can use fink (http://fink.sourceforge.net) to provide complete grass binary packages.
Fink is like the debian package manager (apt-get) and there for very powerfull, as it can check dependencies and download the needed packages too.
Fink already provides 2500 open source packages for MacOSX.
A user would just have to do a
apt-get install grass
And that would download all dependent packates and install them for him.
--
Markus Neteler <neteler@itc.it> http://mpa.itc.it
ITC-irst, Istituto per la Ricerca Scientifica e Tecnologica
MPBA - Predictive Models for Biol. & Environ. Data Analysis
Via Sommarive, 18 - 38050 Povo (Trento), Italy
On Sun, Nov 09, 2003 at 01:27:00PM -0500, Scott W Mitchell wrote:
On Sun, 9 Nov 2003, Michael Barton wrote:
...
> If it would help speed up the release to make it easier for more people
> to test 5.3, I am happy to make the OSX binaries available for
> distribution. I can explain how I configured the compilation, and could
> recompile it if something I did could cause problems for others. I'd
> probably need a bit of help in how to package up the binaries properly
> for distribution. Also, I haven't the foggiest idea of how to make the
> shell script that always seems to accompany binary distributions
...
I'm interested in helping too. I made a binary distribution once before,
and using instructions from Markus, it created that script
"automatically", so that shouldn't be an issue. I'll forward you an email
I have from those days in a minute. In short, there were two ways: (1)
follow the directions in the documentation/release_rules.txt under the
main source directory, and (2) Markus has a script that he uses to get
through all the steps in the release_rules... but unless there are new
developments, this script is not completely portable. On the other hand,
it didn't take TOO much modification to make it work for me on my system,
and it helped make sure I was making binaries the same way as Markus.
The main "problem" with the approach is that it used statically linked
libraries which meant that the binary distribution was HUGE. Markus
recently succeeded to get the CVS version of GRASS to compile with shared
libraries, if I recall correctly.
If I recall correctly, there was a problem in XDRIVER. So I dropped the
idea as 5.3/5.7 without d.mon/XDRIVER is not quite useful.
However, I don't have the information
on how he managed to do this. So, much as I hesitate to bother Markus
with more requests, Markus, if you have more up to date information and/or
scripts that you could share with us from your OS X adventures, can you
pass them on so hopefully we can help keep OS X versions built without as much
intervention from you ?
You can find the magic flags (which took me quite some time to understand
a bit of the MacOSX concept) in GRASS 5.7: