clearlinux / bsdiff Goto Github PK

Hi, I could see the failures in the test script, I suspect the binary loaded in the source got some issue while syncing in the github.
Could you verify and update the test data?

/bsdiff-1.0.4/test$ sh run.sh 
Running test #8 ...
Failed to apply delta (-1)
not ok 4 - 
Running test #10 ...
Failed to apply delta (-1)
diff: 10.out: No such file or directory
not ok 6 - output does not match expected!!

#6 doesn't fully fix the issue.
Consider following files:

ccc.zip

It would be convenient to be able to run the test suite by calling run.sh directly without needing to call make check at toplevel. See #13.

Hello.
Your versions of bsdiff is great with improvements to an original algorithm, and i want to use it on windows too.
It'd be great if some linux specific headers were removed and tool were compatible with MSVC.
Thanks for understanding.

It seems both https://github.com/thoughtpolice/minibsdiff and this project try to do similar things. Is there a chance that the api could be expanded a bit to allow custom compression algos?

After the bspatch program applies a delta to FILE1 to produce FILE2, it runs chmod(2) and chown(2) to set permissions and ownership for FILE2 according to the metadata in the delta file.

This behavior is a convenience for some use cases. For example, consumers of the delta files (like swupd from swupd-client) do not have to bother running the ownership/permission modifications on FILE2 after an update.

But from a testing perspective, requiring root privileges to apply delta files is an inconvenience when those delta files specify to change ownership to an arbitrary user ID and group ID, and the user running the test suite might lack privileges to carry out the operation.

To sidestep the root requirement, it would be nice to either make the chmod(2) and chown(2) calls optional in some way, or add a mechanism to hotpatch delta files on-the-fly to set the user/group ID to the effective user/group ID of the calling process, specifically for the test suite.

There is a disabled functional test (number 14) that suffers from performance issues. The issue needs to be root caused and then we can re-enable that test.

In addition to the current compression algorithms, we should add zstd support.

I have 2 10MB big files where the bsdiff runs 82 minutes
bsdiff file_a file_b delta.diff

files.zip