Well, I don't know the specific case about CoMaps, but forking is a really hard decision, and you need to have strong backup from community to dare going that way. I would assume when projects decide to split is because all the attempts to arrive to a good decision failed. Driving a fork for success is a very hard task, and I guess the majority of forks that fail (a lot) are because they didn't have support from community.
About libreoffice winning over m$ or g$ ? Well, I also would like to see that happening, but not to be replaced by another non-FOSS or half-FOSS option.