[PATCH] D64800: [llvm-readobj] - Stop using precompiled objects in file-headers.test

George Rimar via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Fri Jul 19 10:23:31 PDT 2019


grimar added a comment.

In D64800#1593760 <https://reviews.llvm.org/D64800#1593760>, @sbc100 wrote:

> Is there a danger that by moving in this direction we loose some test coverage since we are no longer verifying the libObject can read a know set of binaries, but only that libObject can read the output yaml2obj?


Having a precompiled binaries committed is very unflexible generally and we always tried to avoid that.
(ok, at least I can say starting from ~2016-2017, I wasn't on LLVM before). Reviewers always asked to use yaml2obj when possible.
Previously yaml2obj was lacked many modern features and many things were impossible to do. Nowadays situation changes every day.

Also, don't forget that we have other projects, like LLD, where libObject based tools like llvm-readobj/objdump check the outputs too.

> i.e. we could evolve the object format in llvm such that these tests would continue pass but the objects would no longer be standards compliant?

At fact many of our tests that use yaml are not standart complaint because of different reasons. I do not think it is a problem, that often helps to
simplify the test case for readability. Sometimes it is impossible to write a test that is standart compliant to test something
(e.g. testing errors in tools like llvm-readelf etc require creating of broken objects).

To summarize: I do not think that often use of yaml2obj is bad. It is a good tool, we might want to improve something,
maybe fixing something, but generally, I think that it produces output no worse than other producers.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D64800/new/

https://reviews.llvm.org/D64800





More information about the llvm-commits mailing list