Hi there ! I've opened a PR on github about non st...
# general
c
Hi there ! I've opened a PR on github about non strict timestamps in zips (https://github.com/pantsbuild/pants/pull/10032). It shows that this is a new parameter in python 3.8, then @hundreds-father-404 suggests to try another method. My second guess about this would be to force pants incremental compilation to update modification date to
now()
even in case of unmodified jar. Do you have any clue for me ?
w
hey @creamy-twilight-68578: responded there, but i suspect that it’s related to the underlying scala code which writes those
c
👍
Hello, I've done some futher tests, and it appear that
org.pantsbuild.zinc.compiler.OutputUtils
may not be responsible of mtime = 0, since creating a
JarOutputStream
and closing it just after creates a jar with correct stat :
Copy code
$ stat foo.jar 
  File: foo.jar
  Size: 22        	Blocks: 8          IO Block: 4096   regular file
Device: fe06h/65030d	Inode: 2384464     Links: 1
Access: (0644/-rw-r--r--)  Uid: ( 1000/mgrenonville)   Gid: ( 1000/mgrenonville)
Access: 2020-06-22 12:01:24.880808759 +0200
Modify: 2020-06-22 12:01:28.464878220 +0200
Change: 2020-06-22 12:01:28.464878220 +0200
 Birth: -