New features in GRASS 5.x - Developers notes

On this page you find latest development news concerning GRASS GIS 5.
You can find descriptions about new features from version upgrades.
Many thanks to all contributors!

[The current development version is available at CVS server]

What's new in GRASS 5 beta9 comparing to beta8

User's convenience

What's new in GRASS 5 beta8 comparing to beta7

Modules

What's new in GRASS 5 beta7/CVS comparing to beta6

Modules

What's new in GRASS 5 beta6 comparing to beta5?

- added missing MAN pages (Markus Neteler)

- d.area: fixed. Now dynamic allocation to remove the 4096 vector
    line limit (Bill Hughes)
- d.legend: fixed bug on shifted numbers. Optional flag to suppress
            display of "no data" now (Markus Neteler)

- i.in.erdas: "-m" reads in map coordinate info (reference) now (Angus Carr)
- i.out.erdas added (Angus Carr)

- m.in.e00: manages correctly the info table with aliases now,
            added a flag for choosing to link geometry and attributes 
            on feature-# or feature-ID, e000b2a integrated (Michel Wurtz)

- nviz code: update prototyping and 'return' fixing (Bill Hughes)

- r.coin fixed (hopefully) (Markus Neteler)

- again general sites import bug (in beta6a-d) fixed
- s.in.ascii fixed for x y z datasets (like DEM) (Markus Neteler)
- s.out.e00 added (Michel Wurtz)
- s.surf.idw import routine updated (Job Spijker, Markus Neteler)
- s.probplt: bugfix on sites read (Job Spijker)
- s.sample fix (Markus Neteler)

- v.bubble: added (Job Spijker)
- v.db.rim, s.db.rim: should compile now
- v.in.arc should be fixed (Werner Droege, Alex Shevlakov)
- v.in.dxf2 added (Ben Johnson)
- v.report: fixed km conversion factor (Markus Neteler)

Shells:
- d.monsize: script to (re-)set GRASS monitor size (Anantha Prasad)
- d.rast.leg2 added (display raster map with legend)
- i.tasscap.tm5: updated for GRASS 5 (Markus Neteler)
- r.reclass.area: updated for GRASS 5 (Markus Neteler)
- slide.show.sh: addec parameter "prefix" to display only selected
                 files (Markus Neteler)

Source code structure:
- New source code tree: src.nonGPL/ (modules not shipped with GPL)
- all libraries under src/libes/ now (Markus Neteler)
- final upates for prototypes, along with several 'return 0;'
   (Bill Hughes)
- updated/checked all Gmakefiles on "mv" command for WIN/Cygnus
   (John Huddleston)
- added HUGE, drand48, srand48 for WIN/Cygnus (John Huddleston)

- configure:
  automatically choosing the diglib64 library at compile-time for 
  64-bit machines (Bill Hughes)

src.related
- rim: should compile now

Many thanks to all contributors!

What's new in GRASS 5 beta5 against former beta4?

added modules
- i.tape.slc (Olaf Hellwich)
- i.in.gtc (Olaf Hellwich)
- i.in.pri (Olaf Hellwich)
- r.out.arc (Markus Neteler)
- r.meta, g.meta, v.meta (Sue Huse?)
- r.sun (Jaro Hofierka)
- r.surf.fractal (Jo Wood)
- r.surf.gauss (Jo Wood)
- r.univar (Markus Neteler)
- s.reclass (Sue Huse) 
- v.plant (S. Cox)

source code:
- 64bit platform detection routine which changes (Luca Palmeri)
   LNG_SIZ from 4 to 8 in case of 64bit architecture
   (file src/mapdev/diglib/portable.h)
- improved prototyping etc. (Bill Hughes)

bugfixes:
- g.region (Luca Palmeri)
- d.colors (cursor movement) (Markus Neteler)
- d.sites (Markus Neteler)
- i.in.erdas (Angus Carr)
- i.oif (Markus Neteler)
- i.rectify (Luca Palmeri)
- i.rectify2 (Luca Palmeri)
- i.maxlik (Angus Carr)
- r.cats (Luca Palmeri)
- r.in.tiff (color table, warning on NULL) (Markus Neteler)
- r.in.gif (color table, warning on NULL)(Markus Neteler)
- r.in.hdf (3.x lib only)
- r.out.hdf (3.x lib only)
- r.answers (Markus Neteler)
- r.mapcalc (Luca Palmeri)
- r.mask (Luca Palmeri)
- r.statistics (Markus Neteler)
- s.surf.rst (Helena Mitasova)
- s.in.ascii (Markus Neteler)
- s.surf.idw (Job Spijker)
- r.resamp.rst (Helena Mitasova)
- v.surf.rst (Helena Mitasova)
- v.digit (Luca Palmeri)
- NVIZ2.2 (Pierre de Mouveaux)
- several code patches from Berhard Reiter for DUAlpha

- shade.clr.sh (Markus Neteler)
- shade.rel.sh (Markus Neteler)

- src/libes/gis/sites.c (Bill Brown): fixes memory problem
- src/libes/imagery/c_point.c (Angus Carr)
- src/libes/geom/optri (Markus Neteler)
- src/libes/g3d/*  (Jaro Hofierka)
- updated around 20 Gmakefiles for SUN/Solaris $(XDRLIB)
- all rst-libs under src/libes/rst/ now
- libimage (NVIZ) under src/libes/libimage/ now

updated man-pages for: (HTML-pages as well) Lisa Zygo 
d.3d                   r.le.dist              r.proj
d.histogram            r.le.null              r.rational.regression
d.leg.thin             r.le.patch             r.report
d.rast                 r.le.pixel             r.spreadpath
g.setproj              r.le.rename            r.stage3
m.in.stf1.db3          r.le.setup             r.stats
p.vrml                 r.le.trace             r.surf.fractal
r.cats                 r.null                 r.surf.gauss
r.circle               r.out.agnps            r.surf.random
r.cost                 r.out.arc              r.surf.xy
r.describe             r.out.ascii            r.tribs
r.in.ascii             r.out.mpeg             r.what
r.in.hdf               r.out.tiff
r.in.pbm               r.plane

Many thanks to all contributors!

From now onwards a bug report formula is provided. If you find a bug in this GRASS 5 release (beta5), please let us know. We try to fix the bugs quickly. Go here for bug report:

http://www.geog.uni-hannover.de/grass/bugtracking/bugreport.html

What's new in GRASS 5.0 beta4 comparing to beta3?

• Lots of bugfixes over 5.0beta3: e.g. fixed "empty text screen" problem, futher improved source code structure, new debug routine G_dump, added missing return statements...
• 121 projections supported now!
• Use m.proj as your "projection calculator" for all 121 projections
  
• DISPLAY: fixed: d.area, d.sites.qual, d.vect.cats
• updated/fixed: g.setproj, m.proj, both support the 121 projections
• RASTER: fixed: r.colors, r.digit, r.flow, r.resamp.rst, r.to.sites, r.slope.aspect, r.contour r.in.poly, r.statistics
• IMAGERY: fixed: i.rectify
• SITES: fixed instable sites format (concerns all site modules)
• fixed: s.surf.rst
• added several sites modules: s.normal, s.perturb, s.qcount, s.probplt, s.rand, s.sample, s.sv, s.univar, s.kcv
• VECTOR: fixed: general bug for all vector modules
• fixed: v.to.rast
• updated: v.in.shape: category text labels support now
• v.digspline renamed to v.digit, removed old v.digit stuff
• added several vector modules from GRASS 4.2.1: v.random, v.extract, v.merge, v.autocorr, v.info, v.export, v.geom
• NVIZ2.2: fixed map selection problem
  - "image dump" writes "xv" readable files now

• Still to do: update r.in.hdf to 4.1.3HDFlib, correct r.surf.contour
• detect further bugs (please report them! or fix them...)
• get vector format 64bit compliant (needs further input)

GRASS 5.x represents the first major change in GRASS GIS functionality in several years.

Please read ChangeLog and CREDITS for authorship.

1. New formats:

2. new applications

3. Improved functionality:

4. New projections:

5. Misc:

      - gets() is an Achilles' heel for GRASS when used on the Web, or where
        through ignorance or malice someone can overrun an input buffer. I 
        have replaced gets() calls with the slightly safer fgets(). 
      - G_warning() and G_fatal_error() are called throughout the GRASS code
        tree as though they accepted variable arguments, i.e. 

           G_warning("Oops: error in %s, i = %d\n",string,integer);

        which is not how they were written. They were written to accept one
        char*, which they write to the screen and/or log. I have re-written 
        these functions so that they accept varargs. This change has not
        been approved by anyone, nor passed any peer review. It just seems 
        right. 
      - G_free() as as companion to G_malloc() and friends. It is annoying
        to have to replace all of the #include <malloc.h> with 
        #include <stdlib.h> so instead I just created a function to be 
        included in the gis.h header. This is portable, which is an 
        improvement. 
    

      - It's much easier to track down the mismatched arguments and implicit
        casts with prototypes. These patches provide prototypes for all of
        the src/libes, and as many other libes and modules as I have gotten
        into. 
      - In the module code, the 4.2.1v21 patch has automatically-created
        header files to provide function prototypes within each module. Some 
        of this is overkill. As I work through each module, these headers
        are fixed to be a little less intrusive. Right now, though, there 
        are still compile warnings that are the fault of this prototyping.
      - The 5.0 code has not had as much work as the 4.2-series, so there
        are still a number of bad calls lurking in the modules. With the 
        proper prototypes, maybe somebody can fix them. 
    

     -  Functions with the wrong number of arguments. 
     -  Functions with the wrong type of arguments. 
     -  Functions with the wrong return type. Usually, this is because a
        function is allowed to default to an 'int' return, even though it 
        returns a char*, or struct*. This may work on 32-bit machines, but 
        whenever sizeof(char*) != sizeof(int) (64-bit machines, for example) 
        these functions overwrite memory, dump cores, and generally fail 
        to work. 
     -  Built-in functions without the proper headers. This leads to bad
        return types. 
  

PhyGeo at University of Hannover, Germany, CAGSR at Baylor University, GMSlabs at University of Illinois and worldwide contributors.
© 1999-2000