rust/tests/run-make-fulldeps/invalid-library/Makefile

include ../tools.mk

all:
	touch $(TMPDIR)/lib.rmeta
	$(AR) crus $(TMPDIR)/libfoo-ffffffff-1.0.rlib $(TMPDIR)/lib.rmeta
	$(RUSTC) foo.rs 2>&1 | $(CGREP) "found invalid metadata"
Diagnose missing includes in run-make tests 2022-08-22 19:00:00 -05:00			`include ../tools.mk`
rustc: Better error when loading invalid libraries When the metadata format changes, old libraries often cause librustc to abort when reading their metadata. This should all change with the introduction of SVH markers, but the loader for crates should gracefully handle libraries without SVH markers still. This commit adds support for tripping fewer assertions when loading libraries by using maybe_get_doc when initially parsing metadata. It's still possible for some libraries to fall through the cracks, but this should deal with a fairly large number of them up front. 2014-03-01 13:09:30 -06:00
			`all:`
rustc_metadata: don't let LLVM confuse rmeta blobs for COFF object files. 2019-11-08 18:06:22 -06:00			`touch $(TMPDIR)/lib.rmeta`
			`$(AR) crus $(TMPDIR)/libfoo-ffffffff-1.0.rlib $(TMPDIR)/lib.rmeta`
Improve error when an .rlib can't be parsed This usually describes either an error in the compiler itself or some sort of IO error. Either way, we should report it to the user rather than just saying "crate not found". This only gives an error if the crate couldn't be loaded at all - if the compiler finds another .rlib or .rmeta file which was valid, it will continue to compile the crate. Example output: ``` error[E0785]: found invalid metadata files for crate `foo` --> bar.rs:3:24 \| 3 \| println!("{}", foo::FOO_11_49[0]); \| ^^^ \| = warning: failed to parse rlib '/home/joshua/test-rustdoc/libfoo.rlib': Invalid archive extended name offset ``` 2021-08-26 13:37:44 -05:00			`$(RUSTC) foo.rs 2>&1 \| $(CGREP) "found invalid metadata"`