Selected: none
Rejected ideas:
swarm - Old, odd and low-performance language (C-Object), decreasing trend. + large user community,
repast - Much fewer user base, lack documentation (tutorials, how-to,
API), java: still not best in performance and difficulties in linking with C libraries + more “modern approach”, used by many computational economists, GIS module,
XML capability by default in java, multiplatform;
Selected: Qt
(+) Include in one framework xml and threads/processes
API, easy to run the same code in Linux/Windows/Mac, lot of doc, tutorials and community;
(-) non-standard make (qmake),
GPL, one-company project,
Rejected ideas:
wxWidgets - Miss some cross-platform libraries other than
GUI (xml, threads/processes), no good ide under windows. +
LGPL, simple to use, “my first love”, community drive project, regmas already implement a simple
GUI in wxWidgets;
GTK - it's a little bit a mess

+ wide usage,
LGPL, community drive project.
Selected: Embedded in QT
Rejected ideas:
xerces-c - Seems to have windows binaryes only for
MS Visual studio. An other requirement. + Olready implementing and working in regmas;
libxml2 - Same as for the GTK.. it seems a little bit a mess for an unexperienced user.
Selected: GLPK
Rejected ideas:
-
COIN-OR - Messy, yet don't understood which is the MIP solver. + They may exist unlinear solvers too.
Selected: direct parsing/writing of GRASS / ArcInfo ascii text grid files.
(+) Not requiring further libraries
(-) Only I/O, not GIS library, only plain ascii grid support
Rejected ideas:
GDAL - Not needed.. only I/O, not a real GIS library, not proved on Windows
Maybe:
SAGA GIS - Untested, project not so large as GDAL + Should be able to perform spacial analisis.. a
real GIS library!
Selected: Umbrello UML
(+) Tested, easy to use, works fine and lot of features (including imports)
(-) On some Linux distros it crashed importing very large projects
Note: give a try also to Bouml