NullPointerException when running bfg -b 10M about bfg-repo-cleaner HOT 6 OPEN

Hi @zhangpy, thanks for sharing this issue with me. Can you tell me which version of The BFG you're using, and are you able to share this repo with me?

from bfg-repo-cleaner.

Can you also give the full command line invocation you used? Were you using --protect-blobs-from?

from bfg-repo-cleaner.

I use version 1.11.6 on ubuntu 14.04-64bit
I tested Java8(oracle 1.8.0_05) and Java7(1.7.0_55-b13)
I am very sorry, I can't share the repo because it's company project.

bfg = java -jar bfg.jar
my command is

cd test.git
bfg -b 10M

I don't use parameter -p.
I tested the parameter and the error message is same.
Please let me know if you need more info.

from bfg-repo-cleaner.

I've had the same issue using --protect-blobs-from, but realised the branches weren't checked out first. Once I checked out the branches, the error disappeared.

from bfg-repo-cleaner.

I'm seeing a similar NPE when using the protect option: here was my command

java -jar ~/Downloads/bfg-1.11.8.jar -b 1M -p HEAD~10 Fusion.git

Is that the correct syntax for -p?

from bfg-repo-cleaner.

java -jar /Downloads/bfg-1.11.8.jar -b 1M -p HEAD10 Fusion.git

The HEAD~10 syntax is valid, and works so long as that commit actually exists in your repo - ie if there actually are 10 commits in the history of your current branch. I've personally tried this on repo, and it works.

However, if there are less than 10 commits in that particular history- ie HEAD~10 does not exist, which you can can test by executing git show HEAD~10 - then the BFG does indeed crash, with an ugly exception stacktrace like this:

[error] (run-main-5) java.lang.NullPointerException
java.lang.NullPointerException
    at org.eclipse.jgit.lib.ObjectIdOwnerMap.get(ObjectIdOwnerMap.java:131)
    at org.eclipse.jgit.revwalk.RevWalk.parseAny(RevWalk.java:807)
    at com.madgag.git.package$RichObjectId.asRevObject(package.scala:190)
    at com.madgag.git.bfg.cleaner.protection.ProtectedObjectCensus$$anonfun$3.apply(ProtectedObjectCensus.scala:66)

...so, sorry the clarity sucks there, but could you verify that that commit actually exists?

Out of curiosity - why have you chosen to protect the contents of an old commit, rather than your current commit?

from bfg-repo-cleaner.