Problem
Upstream purescript-exceptions v6.1.0 exports errorWithName :: String -> String -> Error and the fork does not, so code written against the registry package fails to compile against the fork. Found by the #186 audit's upstream diff; exceptions is otherwise foreign-for-foreign identical to upstream.
Approach
The fork models Error as a plain Lua string: error/errorWithCause return the message, name always answers "Error", and errorWithCause's cause is documented as unobservable through this binding. The minimal restoration follows the same precedent: errorWithName name msg returns msg and the name stays unobservable, documented in the FFI like the cause. If observable names turn out to matter, that is a separate decision to upgrade the whole Error representation (a table carrying name/message/cause), which touches throwException/catchException/showErrorImpl and the Lua-side error/pcall round-trip, and would be its own issue. Add a regression check, release a patch, bump the set.
Prerequisites / Relations
None blocking. Found by the #186 audit's upstream diff (closed). Ships as an exceptions fork patch release plus a package-set bump; an observable-name upgrade of the whole Error representation would be a separate issue.
Verification / Measurement
A program using errorWithName compiles against the fork, and the regression tests pin message (errorWithName n m) == m plus the documented name behaviour.
Problem
Upstream
purescript-exceptionsv6.1.0 exportserrorWithName :: String -> String -> Errorand the fork does not, so code written against the registry package fails to compile against the fork. Found by the #186 audit's upstream diff; exceptions is otherwise foreign-for-foreign identical to upstream.Approach
The fork models
Erroras a plain Lua string:error/errorWithCausereturn the message,namealways answers"Error", anderrorWithCause's cause is documented as unobservable through this binding. The minimal restoration follows the same precedent:errorWithName name msgreturnsmsgand the name stays unobservable, documented in the FFI like the cause. If observable names turn out to matter, that is a separate decision to upgrade the wholeErrorrepresentation (a table carryingname/message/cause), which touchesthrowException/catchException/showErrorImpland the Lua-sideerror/pcallround-trip, and would be its own issue. Add a regression check, release a patch, bump the set.Prerequisites / Relations
None blocking. Found by the #186 audit's upstream diff (closed). Ships as an exceptions fork patch release plus a package-set bump; an observable-name upgrade of the whole
Errorrepresentation would be a separate issue.Verification / Measurement
A program using
errorWithNamecompiles against the fork, and the regression tests pinmessage (errorWithName n m) == mplus the documentednamebehaviour.