It was reported to me that /θ/ no /f/ returns languages with both /θ/ and /f/. The parser builds a stack of queries in order to handle conjunctions; at the end of parsing, all entries on the stack below the first are silently discarded. This is not good - ideally there would at least be an error. But given that the expected behavior was for /θ/ no /f/ to behave identically to /θ/ no /f/ and, it might be better to silently append ands and clear the stack that way. (Maybe display a warning, but we don't have an avenue for warnings, so that would need to be added.)
It was reported to me that
/θ/ no /f/
returns languages with both/θ/
and/f/
. The parser builds a stack of queries in order to handle conjunctions; at the end of parsing, all entries on the stack below the first are silently discarded. This is not good - ideally there would at least be an error. But given that the expected behavior was for/θ/ no /f/
to behave identically to/θ/ no /f/ and
, it might be better to silently appendand
s and clear the stack that way. (Maybe display a warning, but we don't have an avenue for warnings, so that would need to be added.)