Currently, the fuzzer generates pretty much any valid C identifier. I forgot that there are some C identifiers (eg, those with underscores) that Herd can't lex properly.
The easiest solution would be to restrict the fuzzer to generating alphanumeric identifiers. A more sophisticated solution would be to add a mangling step to delitmusification, or switch this restriction off whenever we don't feed the C file through herd (which is the eventual goal).
Currently, the fuzzer generates pretty much any valid C identifier. I forgot that there are some C identifiers (eg, those with underscores) that Herd can't lex properly.
The easiest solution would be to restrict the fuzzer to generating alphanumeric identifiers. A more sophisticated solution would be to add a mangling step to delitmusification, or switch this restriction off whenever we don't feed the C file through
herd
(which is the eventual goal).