A couple of issues (
#26,
#37) have been recently raised against jbovlaste which can be traced to "vlatai", a tool which is used to verify and classify the morphology of new words. vlatai is a tool which is built as part of
jbofi'e: Accordingly it has not been substantially updated for some time, and is known to exhibit bugs, including failing to parse some valid words.
One suggestion has been to replace vlatai with camxes. I've taken some initial steps in that direction, but need some help verifying what the correct behavior should be for the cases where camxes and jbofi'e return different results.
I ran all of the words in jbovlaste through camxes and filed issues
#38,
#39,
#40 and
#41 to record issues encountered with (respectively) cmavo, cmene, fu'ivla and lujvo. I will include a few examples:
- camxes parses {y} as "initialSpaces", but it is considered a cmavo in jbovlaste.
- camxes doesn't parse {bu} compounds like {denpa bu} as cmavo
- {ybu} is a "cmavo cluster" in jbovlaste but a single cmavo per camxes
- {aierne} is a fu'ivla in jbovlaste but a "cmavo + fu'ivla" per camxes
- {selda'ergau} is a fu'ivla in jbovlaste but a lujvo per camxes
- {zei} compounds aren't recognized as lujvo by camxes (nor in vlatai: jbovlaste has a workaround)
If you can offer verifications or corrections for any of these issues, please respond here or add comments to the issues in github.
I'd also like to know if there's consensus on which is more correct or current: "cmavo cluster" (jbovlaste) or "compound cmavo" (CLL), or if there's a distinction between these terms.
Thank you!
--Riley
mi'e la mukti mu'o