Part of ITB 0.1.2 release - See master release

✨ New Features​

Marketplace — browse and install models, template packs & config​

A Marketplace button in the top toolbar opens a catalog dialog where you can browse and install ready-made content into your workspace: whole models, template packs, and config bundles (type & custom-property definitions). Items are organised by tier (Company / Banian / Community) and by group (All / Model / Template / Config), each with a live count, and every item can be expanded to inspect what it contains before you install it.

Show which templates apply to an entity​

A Show templates action on any entity (canvas node, left tree, or right sidebar) previews which template definitions would generate for that entity and lists them in a new Templates tab in the bottom panel. Each match shows its definition, the artifact it produces and the pack it comes from, with per-row "Explain why" (why this template matched) and "View code" (the generated artifact) actions.

Zoom-level attribute detail on large viewpoints​

On large viewpoints the canvas now shows attribute detail only when you are zoomed in enough to read it. Below a configurable zoom threshold, entities collapse to their headers and attribute-mapping edges merge to one line per entity pair; zooming back in restores the full attribute rows and individual connections. A top-left badge indicates when detail is hidden because of the zoom level.

Model-Level Custom Properties​

Custom properties can now be defined and set at the model level, closing the last gap in the typing system's object model — entity, attribute and relation instances already had custom properties; there was no scope for the model as a whole. A new Model category joins the existing three in Settings → Custom Properties: define typed properties there (text or enum, no type-scope narrowing since there are no "model types" to scope to), then set their values once, globally, via Model Actions → Properties.

Key Field on Relation & Viewpoint Types + Type-Key Uniqueness​

Relation types and viewpoint types now carry a key, matching entity types and attribute types — all four type families are consistent again. The Add/Edit dialogs for relation and viewpoint types require a key, and the Settings ▸ Types tables show it as a new column. Alongside this, key uniqueness is now validated for all four type families (previously none of them checked it).

Quality Rules Engine (Model Validation)​

ITB now validates models against declarative quality rules. A new Quality tab in the bottom panel lists the rule groups the user has installed from the marketplace (e.g. the Core Model Quality pack with Referential Integrity, Custom Properties and Model Hygiene) and runs them — all at once, per group, or a single rule — against the current model. While a run is in progress the panel shows live feedback (rules executed out of total, with a per-rule status log). The result is a report with pass/fail/error status per rule, severity counts, and an expandable table of the exact offending records (capped at 100 rows, with the true total shown).

SQL Console Tab​

A new SQL tab in the bottom panel lets power users and support engineers query the current model with real SQL. The model is projected into a read-only, in-browser SQL database (the same engine the Quality tab uses), so joins, aggregations and JSON functions work against familiar table names (entities, attributes, relations, …). The console ships with syntax highlighting, table/column autocompletion, a formatter, a query history (last 10), and a schema reference listing every table and its columns.

Where an installed pack came from​

A pack can reach your installation two ways: through the Marketplace, or as a bundle you upload yourself. Until now that was not visible. Both routes produced the same result on disk, and neither the Marketplace nor the pack dialog could say which one it had been.

a Retail Data Vault demo pack, on the AutomateDV standard macros​

A new template pack, demo_retail_dv, generates a small Raw Vault from three seed files: three staging views, two hubs, one link and one satellite. It is deliberately small, and deliberately not the Banian implementation — its templates call the AutomateDV standard macros (automate_dv.hub, .link, .sat, .stage, v0.11), so the generated models are readable against that package's own documentation.

The client wears the ITB corporate design​

The application's accent colour is now the ITB brand pink, #E9337F, and the ITB logo has replaced the placeholder mark in the sidebar and on the sign-in screen. Until now the client shipped the stock colour of the component library it is built on — a violet nobody chose — so the product looked like its toolkit rather than like itself.

A status bar, and panes you can put away​

The workspace has a strip along its bottom edge holding four toggles — the left sidebar, the panel area, the properties pane, and the console. Each pane can also be shut by a chevron in its own top corner.

The same business world, modelled twice​

The retail demo gains a second, entirely different half. demo_retail_dv already turns customers, products and orders into dbt models for a Data Vault. The new demo_retail_ebo pack takes the same business world and treats it as business objects — producing a JSON Schema per object and an HTML data-catalog page for the whole model.

Bonsai Runs Quality Rules & SQL​

The Bonsai assistant can now validate and query the model the same way a power user does. Three new agent tools were added: run the quality rule groups (all or one) and get a compact pass/fail report with violation samples, run read-only SQL against the model projection, and look up the queryable schema. A findGroups lookup was also added so the assistant can resolve entity-group ids by name.

Wizards show what they will change on existing objects​

The Business Object (BO+) wizard gained a final Review changes step. It lists every change the wizard is about to make to objects that already exist in the model — today the Generate Link+ setting on each link the business object includes — with the current and the new value. Continuing applies those changes together with the business object, as one undoable step.

Column picker grouped by entity, and renaming without the mouse​

The column picker used across the Data-Vault wizards now groups the Available pane by the entity each column belongs to, with a heading per group and its own Add all. In the Business Object wizard, where columns arrive from several satellites at once, that turns a long flat list into one block per source.

🏗️ Infrastructure​

Open documentation with licence badges, and the use case as a dimension​

The documentation site no longer withholds material by licence level — it publishes it and marks it. Every changelog entry carries a badge saying which licence the change needs (Standard or Enterprise), and the release overview groups its counts by licence so the split is visible before you read a single entry.

A pack declares how it is delivered​

A capability pack now states in its own configuration whether it is delivered encrypted or in the open. The field is distribution, it takes exactly two values — encrypted and open — and it is mandatory for a capability pack: there is no default value, and a pack without the field stops the scripts that read it rather than being guessed at in either direction.

One marketplace item installs several components​

The marketplace has a sixth group, Bundles. A bundle is one item that installs several components at once — a model with its type configuration, template packs, capability packs and quality rule groups — in the order that makes them work.

🐞 Fixes​

Large viewpoints render and pan smoothly​

Opening, panning, selecting and expanding entities on large viewpoints (100+ entities, thousands of attribute mappings) is now fast and no longer flickers. Previously such viewpoints could take tens of seconds to appear, stutter badly while panning, and flash the whole canvas when a single node was expanded or selected.

The app becomes usable immediately after loading a large model​

Loading a model (on startup or via reload-from-server) now makes the app interactive right away instead of freezing while the model is written to the local browser cache. On a large model the local write can take many seconds; that work now happens in the background.

Version-drift dialog no longer flashes over the app​

The "version mismatch detected" dialog now opens as a real, standing dialog you resolve deliberately, instead of flashing over the whole app and dismissing itself. It also no longer appears at all during dev render-performance testing.

Smooth canvas gestures on large viewpoints (calm culling)​

Panning and zooming large viewpoints no longer stutters from the canvas engine's own visibility bookkeeping. Rendering off-screen elements is now managed by a "calm" culling scheme that does all its work between gestures instead of on every frame — the worst measured pan gesture is twice as fast, and zooming a large expanded viewpoint up to ~3× faster.

Slim attribute rows (delegated events, no per-cell tooltips)​

Attribute rows in canvas entity nodes are now render-only: row clicks, keyboard activation and the right-click context menu are handled by one delegated handler set per entity table instead of three listeners on every row, the per-cell hover tooltips were removed, and the 🔑 identifier emoji was replaced by a small vector icon. Rows look and behave the same.

Batched attribute-edge layer on large viewpoints​

On large ("heavy") viewpoints, the thousands of attribute-mapping lines between expanded entities are now drawn as a handful of merged vector paths instead of one component per line. The lines look pixel-identical and stay fully interactive (click to select the relation, right-click for its context menu), but panning and zooming an expanded large viewpoint runs essentially zero per-line code — the measured worst case (200 entities, dense mappings) spent 5.4 s of script per pan before, 23 ms after.

Pack settings show readable names instead of internal keys​

When a template pack needs a setting your model does not have yet, the Align model with template pack action creates it for you. Those auto-created settings arrived carrying their internal key as their name, so the properties panel listed entries like gen_link_eff_sat, hub_bk_mode and meta_dt_filter.

Installing a capability pack no longer fails when its base is missing​

A capability pack — the Data Vault wizards, for example — needs the template pack whose vocabulary those wizards create. In the Marketplace the capability pack offered Load even when that base was not installed, and the attempt then failed with a conflict message.

The Quality tab failed once a rule pack was installed​

Installing a quality rule pack from the Marketplace broke the Quality tab: the list of rule sets stopped loading and reported an error instead.

Renaming a relation type no longer switches its templates off​

Relation types are matched by their key now, the way entity and attribute types always were. Rename Business Key Mapping to something that suits your model, and generation keeps working.

Wizards keep your input when you step back​

Every multi-step wizard now redisplays what you already entered when you go back a step, and carries it through to Finish unchanged. Previously a step you returned to came up as if it had never been filled in — and the blank fields it showed were then written over your real answers, so the object was created from the wizard's defaults rather than from what you typed.

Auto-arrange runs again when you pick the same layout​

Clicking a layout in the Auto-Arrange menu now re-runs it, including the one already ticked. Previously, picking a directional layout — Layered or Tree — only opened the Direction row and waited, so choosing the layout you already had did nothing at all.

The attribute collapse arrow on a canvas node points the right way​

The arrow that shows and hides an entity's attributes on the canvas pointed the wrong way round. A collapsed node showed an upward arrow and an expanded node a downward one — the opposite of the collapsible sections in the left sidebar, where a closed section points down and an open one points up.

The relation between two collapsed entities is the one you see​

When two entities on the canvas are collapsed, the line between them is now the entity-to-entity relation. Previously an attribute relation was drawn over it: collapsing an entity merges its attribute lines onto the same route as the entity relation, and the attribute line was painted last, so it covered the relation underneath.

Which packs are installed​

Two places in the client answered the question "what do I have installed?" wrongly, in opposite directions.

Auto Arrange opens with nothing selected​

The Auto Arrange menu on the canvas no longer opens with a layout ticked and a direction highlighted. It opens as a plain list of layouts, and it returns to that state after every run. Picking a directional layout — Layered or Tree — shows the Direction row and waits for you to choose; picking a direction runs it.

A search in the entity tree keeps the folders above its matches​

Searching the left sidebar no longer throws the tree away. A match is shown where it lives — inside its folder, with the folders above it — and folders holding nothing that matches drop out entirely. Search a folder's own name and it comes back whole, with everything in it.

Propagate Attributes finds a path over any modelled relation​

Propagate Attributes Along Lineage now finds a path whenever the two entities are connected at entity level. Previously it only followed relations that already carried column mappings, so a target joined by an ordinary relationship or a dependency reported "No lineage path between these entities" and left the path picker greyed out.

An empty workspace is no longer reported as a version conflict​

Your first edit in a workspace that has never been saved raised "Version mismatch detected" and then failed to save. There was no mismatch — there was simply no model on the server yet. The first edit now takes the lock and the first save creates the model, which is what was always intended.

Installing a capability pack now sets your model up for it​

Installing the Data Vault pack into a fresh model gave you the wizards and nothing to build with — the model had none of the entity types, relation types or custom properties those wizards create. The install reported success, and nothing said otherwise.

Propagate Attributes offers only targets it can actually reach​

The Target entity list in Propagate Attributes Along Lineage now contains only the entities that are actually connected downstream of the source, sorted by name. The lineage path list is sorted shortest route first and names the relation type of every hop. The attribute picker gained Select all, Remove all and a filter, and lists the source entity's columns in the entity's own order.

Part of ITB 0.1.1 release - See master release

✨ New Features​

System Information Sections & License Details​

The System Information dialog (Ctrl + I) now groups version details into clear sections — Client, API, and License — instead of a single flat list. The Client section shows the client version, build, and model version; the API section shows the API version, library version, and a new Build value sourced from the server (shown as "none" until the server reports it). A new License section displays the running instance's licensing: type, who it was issued for, its validity window with a colour-coded status badge, and the version range it covers.

Richer auto-arrange (per-algorithm ELK options, direction & spacing, persisted defaults)​

The canvas Auto Arrange control now configures ELK correctly per algorithm and exposes two new controls:

Hide / show viewpoints & entity groups in the left sidebar​

The left sidebar can get long. You can now declutter it by hiding viewpoints and entity-group folders you don't currently care about, via the right-click menu. Each list (Viewpoints, Entities) gets a small show-hidden toggle in its header that appears only once something is hidden — flip it to reveal hidden items (dimmed, with an eye-off icon) so you can unhide them.

In-app documentation follows the UI language​

The Help menu's Documentation link now opens the user guide in the same language as the app. When the UI is set to German, it opens the German documentation (…/user-docs/de/); in English it opens the English guide (…/user-docs/). The interactive API Documentation (Swagger) link is unchanged.

Wizards can open wide, and picker rows read attribute-first​

Two readability capabilities for any wizard, both driven by a user report that wizard modals were too narrow on large screens and that column-picker rows cut off exactly the text which distinguished them:

Configurable Copy-with-Relation Wizard​

The Copy (original → new) and Copy (new → original) actions now open a short wizard so you can configure the copy before it is created. You can edit the name of the copy, choose whether to create the entity-level and attribute-level relations and which relation type each uses, and choose whether to place the copy on the current viewpoint. Everything is prefilled with sensible defaults, so accepting them reproduces the familiar copy — plus the copy now appears on the canvas.

Create Entities from JSON (Canvas Background Menu)​

Right-clicking the empty canvas now opens a menu with Add entity, Add entity from JSON, and Deselect all. Add entity from JSON (full edition) opens a wizard where you paste a JSON object — or an array of objects — and the modeller creates the matching entities locally. name becomes the entity name, attributes its attributes (a list of names, or objects with datatype/mandatory/…), and every other key becomes a custom property on the entity. New entities are placed on the canvas at the point you clicked.

Duplicate Viewpoint + Duplicate Entity name form​

You can now duplicate a viewpoint: right-click a viewpoint in the left sidebar and choose Duplicate Viewpoint. A short form opens, pre-filled with the viewpoint's name (suffixed (Copy)), description, and type — edit any of them and confirm to create a copy. The new viewpoint shows the same entities in the same layout as the original (the entity boxes are re-placed at their original positions, and their relationships appear automatically).

Import & sync externally-managed data​

The modeller can now import data that is owned by an external source (a dbt project, a Snowflake or Power BI model, …) and keep it in sync. A new Model Actions → External → Import External Data wizard takes an itb-shaped JSON payload, shows a preview of exactly what will change (create / update / delete, with warnings where a change touches your own work), and on confirmation applies it as a single, undoable operation.

Sync relation custom properties + load the payload from a file​

Two Phase-2 additions to Import External Data (ui-050), shipped together:

Propagate attributes along a lineage path​

When a source entity gains a new attribute, you can now carry it downstream to a target entity in one step. Right-click a source entity on the canvas → Propagate Attributes…, pick the target, the lineage path, and the attribute(s) to carry. The tool creates the attribute on each entity along the path where it's missing and wires the attribute-level mapping relations so the column is traceable end-to-end. (Enterprise / full edition.)

External sync adopts objects you already added by hand​

When you import & sync externally-managed data, the sync now reconciles incoming external objects against ones you previously created by hand. If the source starts delivering an object that matches something already in your model (but that wasn't yet managed by that source), the sync adopts it — the existing object is claimed by the source and becomes managed (read-only) from then on — instead of silently creating a duplicate.

Trace an attribute's lineage into a viewpoint​

You can now trace where a specific attribute's data comes from or flows to and see it as a dedicated diagram. Pick an attribute, choose a direction and which relation kinds count, and the tool walks only the relations that attribute participates in — upstream and/or downstream — and generates a freshly named "Attribute Lineage" viewpoint containing just the involved entities, laid out left to right by hop distance. The attribute-level lineage edges render automatically on the new viewpoint. When the viewpoint is created the app immediately switches to it so you can see the result straight away. (Enterprise / full edition.)

Trace an entity's lineage into a viewpoint​

You can now trace where an entity's data comes from or flows to and see it as a dedicated diagram. From any entity, choose a direction (downstream / upstream / both) and which relation kinds count, and the tool walks all of the entity's relations — both attribute-level and entity-level — and generates a freshly named "Entity Lineage" viewpoint of the involved entities, laid out in flow order. The app immediately switches to the new viewpoint on creation. (Enterprise / full edition.)

Bonsai — AI modelling assistant (Review mode)​

A new Bonsai tab in the bottom panel lets you chat with an AI assistant about your model. The first mode is Review: you ask it to critique the model and it streams back a structured, Data Vault / DWH-aware assessment — untyped objects, layer/type/suffix mismatches, malformed hubs/ satellites/links, naming and datatype inconsistencies, orphaned entities, and more. Review is read-only: the assistant can inspect the model but cannot change it.

Bonsai — command flyout & localized command messages​

Typing / in the Bonsai chat now opens a command flyout listing the available chat commands (/clear, /undo, /mode, /help) with a short description of each. The list filters as you type (/u narrows to /undo), and once you type /mode it offers the available modes (Review · Alter · Generate). Choosing an entry — by clicking or with ↑/↓ then Enter/Tab — completes it into the input so you can review (and add an argument) before sending; Esc dismisses the flyout.

Bonsai write modes — Alter and Generate​

Bonsai gained two modes beyond Review, switched with the buttons above the chat or with /mode:

Field-engine default primitives (smarter, safer wizard fields)​

Four generic, descriptor-only enhancements to the shared form/wizard field engine that make prefilled fields both safer and less typing, wired into the Data-Vault wizards (Hub / Satellite / Link / Business Object / Attach-source):

Transfer-list ordering (source-order default, A–Z view toggle, selected-pane reorder)​

Three improvements to the shared dual-pane column picker (the transfer list) used by every Data-Vault wizard step that picks columns — satellite payload, hub / attach-source business keys, link hub slots, and the Business-Object column list:

🏗️ Infrastructure​

lazy elkjs, sync messages in i18n, PocketBase → ITB rename, new field widgets​

Four unrelated cleanups that landed together with the ELK layout variety work:

API-deliverable capability i18n​

Capability labels (button/menu titles, form field labels, placeholders, wizard steps, toast messages) used to live in the client's monolithic messages.ts dictionary, mixed in with app-shell strings. They are now split out so a capability owns its strings as data: every descriptor references its own overlap-free capabilities.<id>.* namespace, and those strings live in per-locale JSON under registry/i18n/{public,enterprise}/{en,de}.json.

transform auto-registration, primitive-coverage diagnostic, declarative source-prefill​

Three capability-architecture improvements toward the "adding a capability = JSON, adding a primitive = a file" goal (retro gap-analysis #3, #4, #6):

🐞 Fixes​

Copy In Between Wizard — Finish Feedback, Configurable Direction, No Discard Prompt​

Three issues raised during testing of the Copy In Between wizard have been fixed:

"Add entity from JSON" — testing-feedback fixes​

A round of fixes to the Add entity from JSON feature based on user testing:

Relationship lines stay aligned on large entities​

Attribute-to-attribute relationship lines on the canvas now connect to the exact attribute row they belong to, regardless of how many attributes an entity has. Previously, on very large entities (e.g. 100+ attributes) the connection points drifted away from their rows and the lines appeared to float below the entity.

cancelling a wizard's "Discard changes?" prompt stopped all wizards from opening​

Cancelling the Discard changes? confirmation on any wizard left the entire wizard system unusable: afterwards no wizard would open again — Add entity from JSON, Copy source to stage, Copy in between, Copy with relation — until the page was reloaded. Ordinary form dialogs (Add entity, Edit entity) were unaffected. The fix re-wires the shared wizard host's open state so the dialog can never desynchronise from the underlying surface.

Fast bulk operations on large entities (delete / duplicate / copy)​

Operations that touch many records at once — deleting, duplicating, or copying an entity together with its attributes, relations, and custom properties — were extremely slow on large entities. Deleting an entity with ~136 attributes and relations took roughly 60 seconds. The same operation now completes in well under a second.

Explain why "Add Property" is disabled when no custom property matches the type​

In the right sidebar's Properties panel (for entities, attributes, and relations), when an object has no custom-property entries the Add Property button is disabled if no custom-property definition applies to that object's type. Previously this looked like a silent, unexplained greyed-out button. It now shows a short message explaining why and where to fix it:

Auto-arrange no longer crashes on self-referencing relations​

The canvas auto-arrange now works on models that contain a self-referencing relation — a relation whose source and target are the same entity. Previously, choosing the Radial (or Tree) layout on such a model silently failed: the nodes never moved and the browser console logged Failed to arrange nodes: RangeError: Maximum call stack size exceeded.

Part of ITB 0.1.0 release - See master release

✨ New Features​

Entity Name Middle Truncation​

Long entity names now display as the first 12 characters + ellipsis + last 12 characters. This keeps the canvas readable without hiding meaningful parts of the name.

Duplicate Entity via Context Menu​

Right-click any entity to duplicate it with all its attributes. The copy is created immediately and can be renamed and repositioned independently.

⚠️ Breaking Changes​

IndexedDB Replaces PocketBase​

The WebApp storage backend has been migrated from PocketBase to browser-native IndexedDB. This change improves offline capability and removes the dependency on an external database process.

Action required: Existing users must perform a one-time model migration after upgrading. Follow the migration guide to export your model from the previous version and re-import it.

Summary​

ITB WebApp 0.1.0 modernizes the storage layer and adds practical shortcuts for day-to-day modeling work. The IndexedDB migration is a breaking change but sets the foundation for better offline support and simpler deployment going forward.

Part of ITB 0.0.8 release - See master release

✨ New Features​

🎯 Advanced Interaction Workflows​

1. Drag Entity from Sidebar to Canvas​

Add entities to viewpoints intuitively by dragging them directly from the left sidebar tree onto the modeling canvas. The grab cursor indicates draggable entities when a viewpoint is active.

Key Benefits:

• More intuitive than context menu workflows
• Entities appear at the exact drop position
• Error handling prevents duplicate additions
• Existing context menu option remains available

2. Drag & Context Menu for Related Entities​

Explore and add related entities directly from the relations panel without manually searching in the sidebar tree. Right-click for context menu or drag to place on canvas.

Features:

• Context menu for quick "Add to viewpoint" / "Remove from viewpoint"
• Drag and drop related entities onto canvas
• Works for all relation categories:
  • Outgoing entity relations
  • Incoming entity relations
  • Outgoing attribute relations
  • Incoming attribute relations
• Grab cursor indicates draggable rows when a viewpoint is active

3. Comprehensive Canvas Context Menus​

Right-click anywhere on the canvas to access context-sensitive actions. Menus are available for:

Entity Node Headers:

• Edit Entity
• Expand / Collapse Attributes
• Remove from Viewpoint
• Copy Source to Stage (ITB API)
• Delete Entity (with confirmation)

Relation Edges:

• Go to Source Entity
• Go to Target Entity
• Edit Relation
• Delete Relation (with confirmation)
• 20px invisible hit area makes thin edges easy to click

Canvas Background:

• Deselect All

All delete actions show confirmation dialogs for safety.

4. User Settings Dialog​

Centralized settings panel accessible via the gear icon in the header bar. All preferences are persisted to localStorage across sessions.

General Tab:

• Language selection (English, German)
• Theme selection (Light, Dark, System)

Canvas Tab:

• Pan Mode - Choose mouse button: Left Click (default), Right Click, or Middle Click
• Background - Dots, Lines, Cross, or None
• Snap to Grid - Toggle with adjustable grid size (10-50px)
• Show Minimap - Toggle minimap overlay

Reset to Defaults button available to restore original settings. All changes take effect immediately.

5. Generate All Artifacts Button​

One-click generation of all artifacts from the header bar. A modal dialog tracks progress through each step: preparing, saving, generating, with success/error feedback.

Workflow:

1. Click the play icon button in header
2. Dialog automatically saves workspace data to backend
3. Backend generates all artifacts
4. Real-time progress indication
5. Auto-dismisses on completion or error

Only visible when backend supports generate-all endpoint.

🎯 Navigation & Discovery​

6. Entity Tree Auto-Reveal on Canvas Selection​

Selecting an entity on the canvas automatically expands all parent folders in the left sidebar entity tree and scrolls the selected entity into view. Keeps sidebar and canvas perfectly in sync.

Behavior:

• Auto-reveal disabled when search filter is active
• No redundant scrolling when selecting from tree directly
• Smooth scroll animation brings entity into view

⚡ Performance Optimization​

7. Comprehensive Canvas Rendering Performance Boost​

Massive performance improvements targeting large viewpoints. Initial render time for ~30 entities with ~40 edges dropped from ~25 seconds to under 10 seconds.

Key Optimizations:

• Viewport Culling - Only visible nodes and edges are mounted in DOM
• Shared Data Maps - Entity types, attributes resolved once at canvas level instead of per-entity/edge
• Structural Equality Guards - Update arrays only when content changes
• Debounced Viewpoint Switch - Node and edge generation batched via requestAnimationFrame
• Pre-resolved Edge Data - All display data precomputed, eliminating per-edge store subscriptions
• Lazy Edge Labels - Labels hidden below 50% zoom
• Cached Handle Positions - Map lookups instead of linear searches
• Inline SVG Icons - Replaced component-based icons with inline SVG
• Singleton Dialogs - One shared AlertDialog and ContextMenu instance
• Optimized CSS Transitions - Scoped properties to avoid full layout recalculations
• Automatic Minimap Optimization - Auto-hides on viewpoints with 50+ entities

Results:

• Faster viewpoint opening
• Smoother panning and zooming
• Faster viewpoint switching
• Significantly reduced jank
• Production-ready performance for real-world data models

Summary​

ITB WebApp 0.0.8 introduces powerful new interaction patterns and user customization options while delivering dramatic performance improvements. Advanced drag-and-drop workflows, contextual menus, and customizable settings enable sophisticated modeling workflows, while canvas rendering optimizations make large-scale data models responsive and usable.

Thank you to all contributors and users who provided feedback!

Part of ITB 0.0.7 release - See master release and API Details

✨ New Features​

UI Enhancements​

This release brings substantial improvements to the user experience with 8 new UI features:

1. Model Filter for Entity Type​

Filter viewpoint displays by entity type to reduce visual complexity. A new funnel button in the model canvas toolbar allows you to toggle visibility of entity types, helping you focus on what matters most.

Key Benefits:

• Improved readability of complex models
• Focus on relevant entity types
• Quick toggle on/off for each type

3. Add Entity to Center of Canvas​

Entities are now added to the center of the visible canvas area rather than at a fixed position, providing a more intuitive editing experience.

4. Entity Usage Indicator Across Viewpoints​

New visual indicator shows where entities are used across different viewpoints, making it easier to understand entity relationships and dependencies throughout your model.

5. Edge Highlighting for Selected Entity​

When selecting an entity, all connected edges are now highlighted, making it easier to trace relationships and understand connections at a glance.

6. Attribute Drawer with Pagination​

The drawer concept has been extended to attributes, now accessible via CTRL+4. All drawers (viewpoints, entities, relations, and attributes) are now paginated, significantly improving performance in large models.

Available Shortcuts:

• CTRL+1 - Viewpoints
• CTRL+2 - Entities
• CTRL+3 - Relations
• CTRL+4 - Attributes (New!)

7. Save Canvas as Image​

Export your viewpoint canvas as a high-quality image for documentation, presentations, or sharing with stakeholders.

8. Auto Arrange Canvas​

Automatically arrange entities in your viewpoint for optimal layout. A layout algorithm now helps organize complex models with minimal manual effort.

9. Resizable Left Sidebar​

The left sidebar can now be resized to accommodate different screen sizes and user preferences, providing more flexibility in workspace organization.

Model Improvements​

10. Mandatory Attribute Field​

Attributes can now be marked as mandatory using a new boolean field. This enhancement provides:

• Clear documentation of required fields
• Better data quality and validation rules
• Support for generating database schemas with NOT NULL constraints
• Backward compatibility (defaults to false)

Display Features:

• Visual indicator * (asterisk) in table columns to mark mandatory fields
• Compact icon-based display saves screen space compared to text labels
• Consistent indicator across main canvas, entity attributes table, and sidebar panels

Example:

{
  "id": "attr_123",
  "name": "customer_id",
  "type": "string",
  "mandatory": true,
  "description": "Unique customer identifier"
}

🐞 Bug Fixes​

2. Model Edge Routing on Collapse​

Fixed an issue where edge routing was not properly updated when collapsing entity nodes in the canvas, ensuring proper visual representation of relationships.

Summary​

Version 0.0.7 represents a significant step forward in usability, performance, and data model capabilities. The UI improvements make working with large models more efficient, while the new mandatory field enhances data modeling best practices.

Thank you to all contributors and users who provided feedback!