Repository navigation
Update jgit from 7.7.1 to 7.8.1 and BouncyCastle from 1.86 to 1.87 - #9795
Conversation
| @@ -1,93 +0,0 @@ | |||
| Name: JGit Library | |||
There was a problem hiding this comment.
todo: investigate why this is not a rename
There was a problem hiding this comment.
already visible in previous PRs https://github.com/apache/netbeans/pull/9576/changes seems to be a git anomaly
There was a problem hiding this comment.
apparently it is a git diff issue triggered by a situation when A is renamed to B and C exists which is identical or very similar to A.
here: some of the licenses were a perfect copy of each other. Fixed it by changing the name field slightly. Future PRs should show it as renamed.
matthiasblaesing
left a comment
There was a problem hiding this comment.
Makes sense to me.
|
Out of interest - what did you observe for bouncy castle 1.87? The modules seem to be launch: and git connection also worked. |
|
i parked the branch at https://github.com/mbien/netbeans/tree/bc187-parked you can see the failures in https://github.com/apache/netbeans/actions/runs/43062157563 |
This won't fly. For example Update: I ran But what I see is, that your HEAD of branch is 1fc7d16, but this is not what was built. That was 6cf72c6:
May the job picked up an incomplete push? |
|
i can take a look at this again tomorrow. Meanwhile i rebased the branch and pushed into mbien#1, this will run the CI in my fork. |
- small adjustments needed due to API changes - added second tripwire test to detect regressions in MAX_REVWALK_FLAGS inconsistencies on jgits side (7.8.0 was broken in that regard) Also update gson from 2.13.2 to 2.14.0 as required by the org.eclipse.jgit.lfs dependency tree. Make the Name headers in the per-module license files module-specific (they were all 'JGit Library' before), so that git's rename detection can unambiguously pair old and new license files in future version bumps instead of arbitrarily cross-pairing them.
|
well. its green. Must have made some mistake, forgot a file or something like that. anyway, the commit is back in this PR like originally intended. CI is running for a final check (on top of the felix update). |
matthiasblaesing
left a comment
There was a problem hiding this comment.
Makes sense to me. Thank you

first:
MAX_REVWALK_FLAGSinconsistencies on jgits side (7.8.0 was broken in that regard but we noticed in time, see Update Bouncy Castle to 1.86 and Apache Mina SSH to 2.20 #9688)second:
very basic manual testing done
note: 8.x would be JDK 25, so lets hope 7.x is maintained for a while longer