datalad/datalad: 1.6.0

Published June 9, 2026 | Version 1.6.0

Software Open

1. Institute of Neuroscience and Medicine, Brain & Behaviour (INM-7), Research Centre Jülich, Jülich, Germany and Institute of Systems Neuroscience, Medical Faculty, Heinrich Heine University Düsseldorf, Düsseldorf, Germany
2. Dartmouth College, Hanover, NH, United States
3. Institute of Neuroscience and Medicine, Brain & Behaviour (INM-7), Research Centre Jülich, Jülich, Germany
4. University of Texas at Austin
5. UC Berkeley - UCSF Graduate Program in Bioengineering
6. UC Berkeley
7. Stanford University, Stanford, CA, United States
8. Psychoinformatics Lab, INM-7, Research Centre Juelich
9. Maze Therapeutics, South San Francisco, CA, United States
10. Potsdam Institute for Climate Impact Research (PIK) e. V.
11. Université catholique de Louvain, Louvain la neuve, Belgium
12. Institute of Climate and Energy Systems - Stratosphere (ICE-4), Research Centre Jülich, Jülich, Germany
13. Eberhard Karls Universität Tübingen
14. Institute of Neuroscience and Medicine, Brain & Behavior (INM-7), Research Center Juelich, Germany
15. Institute of Biophysics, University of Natural Resources and Life Sciences, Vienna, Austria

🚀 Enhancements and New Features

datalad run now wraps command-created commits as merge commits, preserving linear first-parent history while keeping the full provenance record. This works across subdataset hierarchies: each dataset that gained intermediate commits gets its own merge, and merge need propagates bottom-up. A new datalad save --since (since=) parameter makes this independently usable outside of run. PR #7821 (by @yarikoptic)
add submodule URL candidates by URL-rewrite to match parent protocol. PR #7606 (by @bpinsard)

Mark test_gin_cloning as xfail (strict=False) on TimeoutError and IncompleteResultsError instead of flaky retries. gin.g-node.org has become too unreliable from CI runners (HTTP 403 from GitHub-hosted IPs, TCP timeouts, slow responses). When gin cooperates the test still reports XPASS; persistent gin failures no longer block CI. PR #7876 (by @yarikoptic)

Name	Size	Download all
datalad/datalad-1.6.0.zip md5:19931d393c1da269d0aa7bce5bcf0d64	1.8 MB	Preview Download

Is supplement to: Software: https://github.com/datalad/datalad/tree/1.6.0 (URL)

U.S. National Science Foundation
CRCNS US-German Data Sharing: DataGit - converging catalogues, warehouses, and deployment logistics into a federated 'data distribution' 1429999