Sorry I wasn't able to review the attachment in bug #245705 in time. I examined the source and saw that it uses jna to access hunspell library (which it provides a built in version)
So here's what we need to do (some of them being optional tasks)
* I have local ebuilds for jmyspell-core and vldocking which I may commit to java-overlays.
* We already have an ebuild for htmlparser in java-experimental which I may move to java-overlay after testing.
* The source hard depends on both jmyspell-core and jna for the spell-checking but favors one of them at runtime. So here we may choose to patch the sources
to depend on one we choose or ask upstream to make that configurable during build.
* jna in java-overlay depends on gcc-4.3 possibly because the need of libffi. I need someone with gcc-4.3 to do its testing. Other than that I guess we need to patch the sources to look for a system hunspell.
omegat was dropped