Per <>, ...
# pex
Per, I think our bit-for-bit reproducibility of Pex was a lie. Adding
to the integration test, then running
tox -e py37-integration -- -k test_reproducible_build
, results in 4 failures like this:
Copy code
E                       AssertionError: /var/folders/hm/qjjq4w3n0fsb07kp5bxbn8rw0000gn/T/tmplhw07ys7/pex0/.deps/future-0.17.1-py3-none-any.whl/future-0.17.1.dist-info/RECORD and /var/folders/hm/qjjq4w3n0fsb07kp5bxbn8rw0000gn/T/tmplhw07ys7/pex1/.deps/future-0.17.1-py3-none-any.whl/future-0.17.1.dist-info/RECORD have different content.
E                       assert [b'../../bin/...D,,\r\n', ...] == [b'../../bin/f...D,,\r\n', ...]
E                         At index 8 diff: b'future-0.17.1.dist-info/direct_url.json,sha256=7F4J29u5EP7AqxsrmEsapLhIiL0dY8l5GZr40Pzrp_0,229\r\n' != b'future-0.17.1.dist-info/direct_url.json,sha256=lz3ZispHJTXqJGCefZuoDn5ea6grUqSqOiI1savmViE,229\r\n'
cc @enough-analyst-54434 re: our discussion earlier. Lmk if it'd help for me to look more into this. Note the
is what changes
Ok, great. Thanks for digging that up with a repro. I'll look into that and