I still haven't found time to review the many commits in these PRs as higher 
priority things keep coming up.  I'm pretty sure JGit is an issue.  I think 
Wagon as well.  But there could be others.

I don't think reverting is required if it is faster if Carlos and Christofer 
are going to take the time to test and fix the CI release steps against these 
changes.  That's because the nobody except the next RM is currently blocked by 
the changes.  The net result of the CI steps should be the same set of files in 
0.9.6 with different version names.  If you can figure out how to run against 
the 0.9.6 tag you should get the exact same binaries, but if you run against 
the head you'll have some differences against 0.9.6 but if you test local 
builds on Mac and Win you should find any other binary reproducibility issues 
that zlika fixed that Maven may not have picked up.

The argument about changes getting stale are a problem for anyone.  Hopefully 
Christofer is working against a branch and not a fork, and the usual branch 
merging practices need to be used to resync before pushing.

My 2 cents,
-Alex

On 11/19/19, 8:30 AM, "Carlos Rovira" <[email protected]> wrote:

    Hi Piotr,
    
    if you refer to the plan will do, still didn't share. I want to go piece by
    piece to avoid talking of many things at the same time.
    
    I just talked with Chris, and I can revert his PRs, and then we'll
    re-revert again on his repo.
    Or we can left all as is and just restore JGit, if there's nothing more to
    do.
    
    I'll wait to hear from you and Alex to do this.
    Next with that solved, I'll propose a plan to do in Chris's royale fork. At
    that time what I want is people agree that in doing that and making it work
    we'll all be happy and will merge it since it will be something working
    when build and when release. That's the target.
    
    Thanks
    

Reply via email to