Closed jcivlin closed 11 months ago
@dmakarov are there unresolved or more comments?
plz approve if no more comments
End-users won't use the move-mv-llvm-compiler. They'll use
move build
.How debug info is generated when
move build
command is used? How is it tested?
The customer may choose either interface: move build
for package compilation of move-mv-llvm-compiler
for development.
I need this test for development of Dwarf collection in compiler and move-mv--llvm-move
is more convenient for me.
If we need to compare or test how move build
responds to -g
it should be another PR.
End-users won't use the move-mv-llvm-compiler. They'll use
move build
. How debug info is generated whenmove build
command is used? How is it tested?The customer may choose either interface:
move build
for package compilation ofmove-mv-llvm-compiler
for development. I need this test for development of Dwarf collection in compiler andmove-mv--llvm-move
is more convenient for me.
No, we're not going to release move-mv-llvm-compiler to end-users. It's more important that we test the final product that customers will use, than what is more convenient for you.
End-users won't use the move-mv-llvm-compiler. They'll use
move build
. How debug info is generated whenmove build
command is used? How is it tested?The customer may choose either interface:
move build
for package compilation ofmove-mv-llvm-compiler
for development. I need this test for development of Dwarf collection in compiler andmove-mv--llvm-move
is more convenient for me.No, we're not going to release move-mv-llvm-compiler to end-users. It's more important that we test the final product that customers will use, than what is more convenient for you.
I do not know why and when we made this decision. But anyhow this argument is outside of this PR. Plz notice that this PR was already approved by another reviewer.
Thx!
Added dwarf_tests.rs.
Also added macro macro_rules! to_cstring, since usage of the pointer and len of a local CString in a call was incorrect.
Since .dbg_info contains the absolute paths of source names, the content of .dbg_info becomes host sensitive. Test_common.rs was updated with code that helps to mask out this sensitivity before comparison of .actual vs .expected.