What's New This is a big one: two months of daily-driver development, shaped by real translation work and a very sharp beta tester. TRANSLATION MEMORY, GROWN UP - A memory library: one memory per client, shared across every job. Each project writes to its client's memory and reads from as many others as you attach, ranked. - Any memory can be marked down by a percentage, primary included. When a client scores their own TM at 99% on purpose, set the same mark-down and your numbers agree with theirs; docked scores show the mark-down beside them. - View Contents opens any memory in its own window: newest first, searchable as you type, filterable by origin, instant even at a million entries. Correct an entry with the pencil, remove a wrong one for good with the red x, here or in the match panel. - Entries carry their dates: made and last changed. Imported TMX dates are picked up, exported TMX carries them out, and the match panel and concordance show them, so last week's decision reads differently from one inherited years ago. - With multiple translations for one sentence, your own reviewed entries rank first, newest first after that. Correct and confirm a segment and pre-translate picks your version from then on. - Older memories? Recover Entry Dates (Project menu) reads the dates out of previously imported entries in one pass, all memories at once, then removes itself. - Concordance searches every attached memory, shortcut included. PRE-TRANSLATE FROM TM - Fill every unconfirmed segment with a 100% or context match, as drafts for review. Optionally fuzzy matches too, from a slider threshold. - Run one file or the whole project: every file fills, opened or not, saving as they complete, repeated sentences looked up once. - Filled segments show their score in the Match column; the sidebar's 100%, Fuzzy, MT and No match views sort by leverage. REVIEW JOBS - Comment on any segment (right-click or shortcut) as a comment, correction or question. Comments never end up in the delivery file. The QA Report grows a Comments category; on a review job the translator's original wording shows struck through above your edit. Untick the other checks and the HTML export becomes a clean Comment Report to send along. - Comments coming back import the same way as LinguaReview's files, automatically when the file sits next to the document. QA REPORT - One sweep across the whole project: the editor's checks plus consistency both ways, same source translated differently and same translation reused, across all files together. - Findings per check with clickable segment numbers; comments survive re-runs. Which checks run and which files count is remembered. - Export as HTML for a client-ready report. Dark mode on screen, always prints light. GLOSSARIES - In the library too: per client, ranked, same View Contents window. - Merging shows what would change first; a pure duplicate offers to remove itself. - Term matching understands inflected forms, so a plural never hides a term, and treats all-caps codes strictly, so DI or E never lights up ordinary words. IN THE EDITOR - Show Invisibles: spaces, tabs, breaks and non-breaking spaces made visible, stowaways like zero-width spaces flagged in coral. Purely visual. Non-breaking space in the Edit menu, shortcut included. - Empty tag pairs show as one chip; double-click to type inside. Inserting either half brings the other along, or select text and click a pair tag to wrap it. Tags drag from source to target. - Find and Replace narrows the grid to matches, tints every hit; Replace All undoes in one step. - Check consistency sweeps a whole multi-file session in one pass. - Right-click a segment for confirm, comment, review and lock; a multi-selection acts as one. - Text size controls, Control-Tab file switching, reference windows that float over full screen.

What's New This is a big one: two months of daily-driver development, shaped by real translation work and a very sharp beta tester. TRANSLATION MEMORY, GROWN UP - A memory library: one memory per client, shared across every job. Each project writes to its client's memory and reads from as many others as you attach, ranked. - Any memory can be marked down by a percentage, primary included. When a client scores their own TM at 99% on purpose, set the same mark-down and your numbers agree with theirs; docked scores show the mark-down beside them. - View Contents opens any memory in its own window: newest first, searchable as you type, filterable by origin, instant even at a million entries. Correct an entry with the pencil, remove a wrong one for good with the red x, here or in the match panel. - Entries carry their dates: made and last changed. Imported TMX dates are picked up, exported TMX carries them out, and the match panel and concordance show them, so last week's decision reads differently from one inherited years ago. - With multiple translations for one sentence, your own reviewed entries rank first, newest first after that. Correct and confirm a segment and pre-translate picks your version from then on. - Older memories? Recover Entry Dates (Project menu) reads the dates out of previously imported entries in one pass, all memories at once, then removes itself. - Concordance searches every attached memory, shortcut included. PRE-TRANSLATE FROM TM - Fill every unconfirmed segment with a 100% or context match, as drafts for review. Optionally fuzzy matches too, from a slider threshold. - Run one file or the whole project: every file fills, opened or not, saving as they complete, repeated sentences looked up once. - Filled segments show their score in the Match column; the sidebar's 100%, Fuzzy, MT and No match views sort by leverage. REVIEW JOBS - Comment on any segment (right-click or shortcut) as a comment, correction or question. Comments never end up in the delivery file. The QA Report grows a Comments category; on a review job the translator's original wording shows struck through above your edit. Untick the other checks and the HTML export becomes a clean Comment Report to send along. - Comments coming back import the same way as LinguaReview's files, automatically when the file sits next to the document. QA REPORT - One sweep across the whole project: the editor's checks plus consistency both ways, same source translated differently and same translation reused, across all files together. - Findings per check with clickable segment numbers; comments survive re-runs. Which checks run and which files count is remembered. - Export as HTML for a client-ready report. Dark mode on screen, always prints light. GLOSSARIES - In the library too: per client, ranked, same View Contents window. - Merging shows what would change first; a pure duplicate offers to remove itself. - Term matching understands inflected forms, so a plural never hides a term, and treats all-caps codes strictly, so DI or E never lights up ordinary words. IN THE EDITOR - Show Invisibles: spaces, tabs, breaks and non-breaking spaces made visible, stowaways like zero-width spaces flagged in coral. Purely visual. Non-breaking space in the Edit menu, shortcut included. - Empty tag pairs show as one chip; double-click to type inside. Inserting either half brings the other along, or select text and click a pair tag to wrap it. Tags drag from source to target. - Find and Replace narrows the grid to matches, tints every hit; Replace All undoes in one step. - Check consistency sweeps a whole multi-file session in one pass. - Right-click a segment for confirm, comment, review and lock; a multi-selection acts as one. - Text size controls, Control-Tab file switching, reference windows that float over full screen.

Bug fixes and improvements Added a new QA check: "Invalid Tags" in the QA section. Segments with invalid tags can't be confirmed until the tags are fixed, since broken tags cause import issues in other CAT tools. Added support for Phrase files.

Bug fixes, (QA) improvements