v4.4.0

Added

• Add operator negation with existance assumption to AQL
  by upgrading to graphANNIS 1.1.0

v4.3.1

Fixed

• Fetching the segmentation span in CSVExporter could fail if only one token was covered.

v4.3.0

Added

• Support the segmentation parameter in the CSVExporter. This allows to get the spanned text
  not from the token (which can be empty), but from the given segmentation layer.

Fixed

• User guide was still mentioning the non-existing WekaExporter

v4.2.0

Changed

• Always show the selected corpora at the beginning of the grid, even when not included in the
  current filter. This should make it much easier to spot corpora that where selected by mistake.
  (also fixes issue #563).

Fixed

• Exception thrown when URL with corpus fragment (#c=...) was openend
• Fix width of the corpus list columns when scrolling to long corpus names
• Info and browse document buttons were not always visible without scrolling
• Removed internal usage of the clone() function which has been reported
  as "blocker" code smell by the sonarcloud static code analysis.

4.1.4

Fixed

• Matches for documents or (sub-) corpora where not shown. There is a special SingleCorpusResultPanel which should have been used, but was not triggered.
• Media player was not loaded in result view
• Security update in dependency jsoup (from 1.11.2 to 1.14.2)

4.1.3

Fixed

• Query result order was not considered in result view

4.1.2

Fixed

• Use updated graphANNIS version 1.0.2 to fix issues in performance of subgraph extraction
• Instance configurations where not loaded (including the default one)

4.1.1

Fixed

• Do not use the default read timeout in the HTTP REST client (#712)
• Use an GitHub Action that only uploads the release files instead of creating a new release
• Download the test corpora before building the release artifacts

4.1.0

Added

• Support for Oauth2 authorization services like https://github.com/korpling/shibboleth-oauth2-forwarding instead of only OpenID Connect

Changed

• Release binaries are generated by GitHub Actions and added to GitHub Releases
  (instead of releasing ANNIS on Maven Central)

Fixed

• Editing a group shows an error message about a missing
  PopupTwinColumnSelectMultiline class.
• Default server configuration now creates the needed SQL tables for the reference
  links. This fixes an error message in the administrator view.
• Removed instable inclusion of CITATION.cff file in "About ANNIS" window.

4.0.0

This release is a complete overhaul of the ANNIS service backend. Instead of using the relational database PostgreSQL, a custom AQL implementation based on graphs called graphANNIS is used. This stable release marks an end to an extensive phase of beta-testing.

Compared to the last beta release, the following issues have been fixed

Fixed

• Do not discard dominance edge, when only the target node has an incoming unnamed incoming edge.
  For the GUM corpus with mixed dominance for syntax and RST trees, this caused some segments to
  have no connection to any token. (#696)
• Fix display of document meta annotations in the document browser
• Desktop UI aborted execution on KDE when trying to open the browser (#702)