github.com/treeverse/lakefs@v1.24.1-0.20240520134607-95648127bfb0/clients/hadoopfs/CHANGELOG.md (about) 1 # Changelog 2 3 ## _Upcoming_ 4 5 ## 0.2.4 6 7 lakeFSFS: new Token Provider feature with IAM Role Support for lakeFS authentication (#7659 + #7604) 8 9 ## 0.2.3 10 11 * Fix createDirectoryMarkerIfNotExists (#7510) 12 13 ## 0.2.2 14 15 * Update lakeFS SDK to 1.12.0 16 * Use overwrite in lakeFSLinker to control If-None-Match 17 18 ## 0.2.1 19 20 * Update lakeFS SDK to 1.0.0 21 22 ## 0.2.0 23 24 ### Breaking changes 25 26 * New required fields added to StorageConfig API (#6509). **:warning: This 27 is a breaking change:** Only lakeFS server versions >= 0.108.0 are 28 supported. 29 30 ### Bug fixes 31 32 * Better handling of quotes and encoding of ETags / checksums, especially 33 relevant to Azure (#6756). 34 * Correctly parse physical URIs that contain whitespace (#5827). 35 36 ### This release contains significant maintenance changes 37 38 None of these are user-visible, but all may be of interest. 39 40 * Unit test using mockserver (#6646). Unit tests now more accurate _and_ 41 faster. 42 * Use upcoming future-proof SDK (#6737). This change servers as an example 43 of how you might transition other code to the new future-proof SDK. 44 45 Additionally, :warning: if you do not use an assembled hadoopfs "Über" Jar 46 then you will need to change your dependencies to include `io.lakefs:sdk` 47 rather than `io.lakefs:api-client`. 48 49 ## 0.1.15 50 51 * Fix presigned mode on Azure fails getting blockstore type (#6028) 52 53 ## 0.1.14 54 55 * Fix bug #5603: fix issues with Delta tables on Databricks 9.1 ML. 56 * Reduce listObjects calls and results amount (#5683) 57 58 ## 0.1.13 59 60 * Presigned mode <sup>BETA</sup>: a secure, cloud-agnostic way to use lakeFS with Hadoop. 61 62 ## 0.1.12 63 64 Notable bugfixes: 65 * Fixing CopyObject -> StageObject fallback mechanism to support old lakeFS versions 66 67 ## 0.1.11 68 69 What's new: 70 * Upgrade lakeFS client to v0.91.0 71 * Prefer to use CopyObject instead of StageObject if possible in rename operation. 72 73 ## 0.1.10 74 75 * Performs fewer API calls, leading to some performance improvement. 76 * More consistent shading: everything now under io.lakefs.hadoopfs.shade 77 78 Notable bugfixes: 79 * Don't link twice when close() is called twice after create() -- support 80 lakeFS KV with uncommitted GC. 81 * Make Hadoop a provided dependency, improving compatibility with some 82 existing Spark configurations. 83 84 ## 0.1.9 85 86 Bump jackson-databind library dependency. 87 88 ## 0.1.8 89 90 The FileSystem implementation can now work with any Hadoop version. 91 92 To use it, the classpath needs to include the _hadoop-common_ and _hadoop-aws_ packages, including and all their dependencies. 93 94 ## 0.1.7 95 96 ## Performance: 97 98 * Speed up recursive delete by using deleteObjects.