Hi, I've seen this pull request that have been merged : https://github.com/treeverse/lakeFS/pull/1310 But I do not understand how i can disallow empty commit on a repo. Is this possible?

Hi @Samuel THIRIET after the above PR you can't commit without changes. If you try, for example, to use lakectl and commit without don't any changes to your branch - the commit will fail the commit.

I have made more tests : The failure occured when there is no change but there is no failure, for example, if i upload the same file a second time and try to commit :

root@74f1071332bb:/opt/bitnami/spark# lakectl fs upload <lakefs://my-repo/main/imdb/title.episode.tsv.gz> --source title.episode.tsv.gz
Path: imdb/title.episode.tsv.gz
Modified Time: 2021-11-30 13:04:33 +0000 UTC
Size: 28145618 bytes
Human Size: 28.1 MB
Physical Address: <local://storage-location/a1fd8e689f794ef39939c074b282242a>
Checksum: 990ca31cfa7cfecd36badecf08a4dec7
root@74f1071332bb:/opt/bitnami/spark# lakectl commit <lakefs://my-repo/main> --message "first commit"
Branch: <lakefs://my-repo/main>
Commit for branch "main" completed.

ID: 326bd71be9b17e09edbbf988c4a7ccf1253cfe6253df29231bd2e27b1daed603
Message: first commit
Timestamp: 2021-11-30 13:05:25 +0000 UTC
Parents: f742d9bb095fb689489cf7a2735ea7cfe7024513d648aae613bbcf7c23aa731d
root@74f1071332bb:/opt/bitnami/spark# lakectl fs upload <lakefs://my-repo/main/imdb/title.episode.tsv.gz> --source title.episode.tsv.gz
Path: imdb/title.episode.tsv.gz
Modified Time: 2021-11-30 13:06:44 +0000 UTC
Size: 28145618 bytes
Human Size: 28.1 MB
Physical Address: <local://storage-location/202bfa1d035f442685709f223f195999>
Checksum: 990ca31cfa7cfecd36badecf08a4dec7
root@74f1071332bb:/opt/bitnami/spark# lakectl commit <lakefs://my-repo/main> --message "second commit"
Branch: <lakefs://my-repo/main>
Commit for branch "main" completed.

ID: 5ad9c4118406bbdd1bee1452367c25f9b45505d15c15fcbc5541aad09f7f2fec
Message: second commit
Timestamp: 2021-11-30 13:06:49 +0000 UTC
Parents: 326bd71be9b17e09edbbf988c4a7ccf1253cfe6253df29231bd2e27b1daed603
root@74f1071332bb:/opt/bitnami/spark# lakectl diff <lakefs://my-repo/5ad9c4118406bbdd1bee1452367c25f9b45505d15c15fcbc5541aad09f7f2fec> <lakefs://my-repo/326bd71be9b17e09edbbf988c4a7ccf1253cfe6253df29231bd2e27b1daed603>
Left ref: <lakefs://my-repo/5ad9c4118406bbdd1bee1452367c25f9b45505d15c15fcbc5541aad09f7f2fec>
Right ref: <lakefs://my-repo/326bd71be9b17e09edbbf988c4a7ccf1253cfe6253df29231bd2e27b1daed603>

just to verify - when you upload the same file the second time. it is listed as a change as the timestamp was updated, right?

Yes it is listed as changed before i commit (because of the modification date metadata?)

root@74f1071332bb:/opt/bitnami/spark# lakectl fs upload <lakefs://my-repo/main/imdb/title.episode.tsv.gz> --source title.episode.tsv.gzPath: imdb/title.episode.tsv.gz
Modified Time: 2021-11-30 13:19:48 +0000 UTC
Size: 28145618 bytes
Human Size: 28.1 MB
Physical Address: <local://storage-location/e2808480d77b4a8f8956f7b6dd4c2c80>
Checksum: 990ca31cfa7cfecd36badecf08a4dec7

root@74f1071332bb:/opt/bitnami/spark# lakectl diff <lakefs://my-repo/main>                                                        
Ref: <lakefs://my-repo/main>
~ modified imdb/title.episode.tsv.gz

yes

yes exactly

how do you see your workflow for this? part of the commit? or part of a flag when you upload the data? because we will keep the old metadata.

perhaps on the upload action? by checking the checksum before the upload?

Skip upload if the content checksum is the same as a lakectl option? Note that if you still upload a file using our S3 interface or API in this case a change will be made. How do you use lakefs? If the change is requested from different clients, the above improvement will not be helpful.

yes you're right. it's not very consistent

Thanks I'll open it soon