"Tobias C. Rittweiler" <tcr@...> writes:
> SUB-COMPILE-FILE in src/compiler/main.lisp catches all
> FATAL-COMPILER-ERRORs signalled during the compilation attempt.
> It then SIGNALs the caught condition again. I wondered about why it
> doesn't use ERROR in that case. Perhaps it's not allowed by the
> standard which seems to talk about signalling errors only.
> (There's also the comment in compiler/compiler-error.lisp which states a
> policy that COMPILE-FILE should not invoke the debugger for users'
> errors. [*])
After having slept, this all is a bit less confusing to me now, and
actually makes sense. COMPILE-FILE should indeed not throw us into a
> But what I find perplexing is the fact that it'll print some error
> output after the signalling. So it seems that people _have to_ know
> about SB-C:FATAL-COMPILER-ERROR and have to catch it if they invoke
> COMPILE-FILE, and want to avoid any possible printage. Is this really
> intended, or am I simply missing some piece of the puzzle?
This makes in so far sense as it doesn't seem to be possible to have
COMPILE-FILE not print anything even in the normal ERROR or WARNING