Wrong parsing of "ローマじゃない"

[nicolasmaia]

When parsing ローマじゃないです, Rikaichamp thinks it's looking at ローマ字. This problem can maybe be fixed if you tell the program to never separate じ from ゃ (and other similar cases).

[birtles]

That's a good idea!

[nicolasmaia]

This seems fixed from my end. Did you tweak with it?

[birtles]

Oh, no I don't know what happened there. I mean, I half rewrote the de-inflection code so that probably fixed it but actually I didn't expect the behavior to change so I'm afraid I might have broken other things in the process.

[birtles]

git bisect says this was fixed by fa1c7f7 but I have no idea why.

[birtles]

I'm closing this for now, but I really don't know exactly what fixed it.

[neneko]

This seems to be broken again.

[birtles]

Should be fixed in the next version.