New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
save would make git crash replacing a file with directory with content #6558
Labels
Comments
update: Junio (of git) already replied confirming it is a bug in git (citing: "guess is wrong and it is a bug that it does not notice it") and (rightfully so) reprimanded me for saying that |
yarikoptic
added a commit
to yarikoptic/datalad
that referenced
this issue
Mar 18, 2022
accompanied with a test to demonstrate save defect on renaming file into directory. Prior Commits optimizing the body/data structures of _save should have helped to also minimize potential run time for this added check/manipulation of status. Underlying issue is actually within git, see datalad#6558 and correspondence on git mailing list https://lore.kernel.org/git/xmqqv8wdk308.fsf@gitster.g/T/#madd83451d39754c044691910f3897659ed291abf
yarikoptic
added a commit
to yarikoptic/datalad
that referenced
this issue
Mar 18, 2022
accompanied with a test to demonstrate save defect on renaming file into directory. Prior Commits optimizing the body/data structures of _save should have helped to also minimize potential run time for this added check/manipulation of status. Underlying issue is actually within git, see datalad#6558 and correspondence on git mailing list https://lore.kernel.org/git/xmqqv8wdk308.fsf@gitster.g/T/#madd83451d39754c044691910f3897659ed291abf
This was referenced Mar 21, 2022
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
might populate with more details later, for now just a record of reproducing the problem reported in dandi/dandisets#127 (comment) and shown in CI runs of dandi/dandisets#134
cut paste of manual reproducing
which boils down to "not provide path of removed (and staged for removal) directory into 'git commit' call"
DataLad 0.15.5 WTF (configuration, credentials, datalad, dataset, dependencies, environment, extensions, git-annex, location, metadata_extractors, metadata_indexers, python, system)
WTF
configuration <SENSITIVE, report disabled by configuration>
credentials
datalad
dataset
dependencies
environment
extensions
git-annex
location
metadata_extractors
metadata_indexers
python
system
The same problem with bleeding edge master datalad and most recent git available in debian ATM
I think we should fix it in DataLad via "remove paths from the call to
commit
which are parents for another path in the call" but that might be too expensive although I remember we had already done something like that for something else.but I also think it is worthwhile reporting to git folks -- IMHO operation is inconsistent
e.g. if I remove the file and regardless either I add that to index or not -- commit works fine
The text was updated successfully, but these errors were encountered: