Skip to content

v5.0.0

Latest

Choose a tag to compare

@riccio82 riccio82 released this 10 Nov 10:24
· 172 commits to develop since this release

Matecat 5.0 is a major release: a refreshed interface across the whole app, new project creation and editor features, a broad security hardening pass, and many reliability fixes in analysis, review and search.

✨ Highlights

  • More pre-confirmation control: pre-confirmation rules can now include 101% matches and set the exact state they get in the editor. Pre-confirmation and variant-matching settings moved to the new Advanced sub-tab of the translation memory settings.
  • XLIFF import settings: decide whether pre-populated XLIFF segments are locked in the editor.
  • Redesigned Projects page: a new layout for project cards and their jobs, with sticky project and chunk headers, a "View more" toggle for long job lists, and refreshed team and member management.
  • Segment history in the quality report: open a segment's history to see every confirmation step and the exact text confirmed at each stage.
  • Context preview: can open automatically, stays pinned to the bottom of the page, maps AEM segments that are a single sentence of a block, and shows placeholders as text.

🖊️ Editor

  • Search and replace: replace all can include locked segments; replacement is applied only after the segment save succeeds; replacing in a locked segment shows the locked alert; replace all no longer breaks version history, undo, reviewed word counts or LQA issues.
  • Compress and expand toggle for ph tags, now also for g and x tags carrying equiv-text; XLIFF 2.0 inline codes show their disp/equiv value.
  • All locked segments are now out of scope, not only locked 101% in-context matches. This applies to the reviewed word count, QA issue creation, autopropagation, search and the segment filter.
  • QA warning when a fuzzy match is confirmed without changes, and fuzzy match differences are highlighted again.
  • Reference-language matches can be picked independently, and the two reference languages can no longer be the same.
  • Job progress bars show the correct stacked percentages for translation, revision and revision 2.
  • Sticky file bar showing the current file name, which blinks when the file changes off screen.
  • Live updates across browser tabs when a segment is disabled or enabled.
  • Bulk selection bar that confirms bulk status changes, and animated search, filters and QA submenus.
  • Termbase footer messages are shown as toast notifications.
  • lexiQA spellcheck suggestions with inline word replacement; unsupported locales now fall back to a supported lexiQA language.

📁 Projects, upload and analysis

  • Translation memory settings split into Resource selection and Advanced sub-tabs.
  • Mandatory issue marking: choose which review rounds (R1, R2, both or none) require the reviewer to add an issue before approving a segment they edited. The choice can be saved in project templates.
  • Fixes for projects stuck in analysis, stranded multi-language TM analysis and word counts of placeholder-only segments.
  • MTQE (QE workflow) projects: volume analysis fixed and job summary cards styled.
  • Job metadata is carried across split, merge and password rotation; concurrent job splits no longer deadlock.
  • An error is shown when generating a revise 2 link fails, instead of failing silently.
  • EPUB is now a supported file format; reorganised supported formats list.
  • Clearer 100-file upload limit, with a tooltip, a scrolling list and a counter.
  • ICU segments are subfiltered with ICU-compliant handlers in analysis and contributions.
  • New filter options: inner_content_type for JSON and force_double_quoting for YAML.

🤖 Machine translation and TM

  • Lara receives multiline text when the project uses paragraph segmentation.
  • MyMemory receives the owner's job keys as mt_key, and writes answered with a non-JSON body are requeued.
  • Glossary updates no longer show a misleading error message.
  • AI assistant defaults to the gpt-5.6-luna OpenAI model, follows the current OpenAI API and reports errors safely.
  • Queue messages are retried with exponential backoff.

🎨 Design

  • Redesigned Projects page (Manage): a new layout for project cards and their job and chunk rows, and restyled team and member management.
  • Refreshed design system: redesigned select and dropdown, and restyled status, error and maintenance pages.
  • All app copy (UI labels, error messages and emails) is now in sentence case.
  • A proper 404 page for unknown addresses.
  • Design review fixes across the editor, volume analysis, quality report, settings, activity log and modals.

🔒 Security

  • Authorization hardening across jobs, projects, splits, templates, activity log and volume analysis (IDOR fixes).
  • Session and login token lifecycle hardened: every token is revoked on password change and reset, and auth tokens are scoped to the flow that issued them.
  • Upload token validation, lower upload limits and a hardened PHP configuration.
  • CORS, XSRF, CSP and .htaccess hardening; plugins can add their own CSP sources.
  • User-typed names (teams, projects, keys) are normalised, bounded and escaped.

🛠️ Platform and API

  • v1/new no longer requires payable_rate_template_name, and omitted optional template fields default instead of returning a 500.
  • Several production crashes reported by Sentry are fixed.
  • GDPR user:gdpr-anonymize command and a database:clean job and project erasure command.
  • Cache eviction now follows database transactions, so stale rows are no longer cached after a commit or rollback.
  • Swagger documentation for the segment-analysis endpoints, lara_style and the CLDR plural rules of the v3 languages endpoint.
  • The whole frontend is now function components, and the PHP codebase passes static analysis with no baseline.

Full changelog: v4.0.0...v5.0.0