Closed goodmami closed 4 years ago
The problem can be seen from the first few lines of an AVM in the file:
qc_unif_trad_pack := *top* &
[ ARGS.SYNSEM.LOCAL.CAT.HEAD "0" #| 5.1e+05 |#,
ARGS.SYNSEM.LOCAL.CAT.VAL.COMPS "1" #| 2.7e+05 |#,
ARGS.SYNSEM.PUNCT.RPUNCT "2" #| 1.5e+05 |#,
ARGS.SYNSEM.LOCAL.CAT.HEAD.MOD "3" #| 1.2e+05 |#
[...]
The value of ARGS.SYNSEM.LOCAL.CAT.HEAD
is the string "0"
, but in the fourth feature path it is treated as a sub-AVM in ARGS.SYNSEM.LOCAL.CAT.HEAD.MOD
. These wouldn't unify, so I think this file is processed specially by PET.
I think an appropriate fix, then, is to return a more useful error message.
I'm getting an error parsing the pet/qc.tdl file of the ERG, and the error message isn't terribly indicative of the problem:
If there's a problem with the TDL there should be a TDLSyntaxError.