md5sum hashes do not match when you build the same code twice on macos about rules_scala HOT 7 CLOSED

good catch. Would be willing to send a test that checks this and a PR to fix it?

from rules_scala.

actually, I think I can fix it while I'm looking at it.

from rules_scala.

actually, can you post how you solved this exactly? my attempts are not working with bazel 0.2.3.

I can't create the manifest in the jar. Can you take a look at #55 and share your solution in a PR?

from rules_scala.

oh we're not using 0.2.3 yet. we are still using 0.2.1 right now because 0.2.3 seems like it was just broken. Might move to 0.2.2 soon, though. Let me take a look, though. Basically we're not using "jar" to jar it. Using zip with the -FS option, it definitely keeps the timestamps.

from rules_scala.

0.2.3 broke the scala rule with changes but they have now been fixed.

Can you send a PR to the repo with your change to scala.bzl? Maybe I just
didn't get the right zip flags to make it work.

On Wednesday, May 18, 2016, xynny [email protected] wrote:

oh we're not using 0.2.3 yet. we are using 0.2.1 right now because 0.2.3
seems like it was just broken.

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub
#53 (comment)

P. Oscar Boykin, Ph.D. | http://twitter.com/posco | http://pobox.com/~boykin

from rules_scala.

so I don't know if i can submit a PR because of the cla thing, but you can see this one i just changed now:

https://github.com/xynny/rules_scala/blob/master/scala/scala.bzl#L141-L147
https://github.com/xynny/rules_scala/blob/master/scala/scala.bzl#L51

Those are the two things that made the hash stick for most of the created jars. For some reason I'm still getting a different hash for HelloLib.jar. the thrift.bzl rule still has the jar command that creates the different hashes, though. Basically we switched to the zip method in all of our rules and now local compile doesn't break cache. I looked into rewriting this in code with the java zip util, too, because you can reset metadata, but since this worked I haven't done it.

from rules_scala.

this was fixed by #55

from rules_scala.