For users upgrading from 15.0 (September 2, 2026) to 15.1

15.1 follows 15.0 by two weeks and resolves 55 tickets. Most of the work went into three areas: populations that can say more, importing models from XML Schemas, and day-to-day robustness in the modeler and the merge. This page covers what changes for you. The full per-ticket changelog ships with the installer.

1. Populations that keep order and group tuples

The population editor grew the most in this release.

Ordered populations (#5993) — tuples of a fact type keep the order you give them. Arrange rows in the editor; verbalization, diagrams, the expression viewer and the MCP tools all walk the tuples in that order. There is no separate 'ordered list' mode any more: arranging rows is always allowed.
Tuple groups: sets, bags and sequences — a role now says what it holds. A role can hold a single value, a set, a bag or a sequence, and the diagram brackets the role number accordingly. In the editor you group and ungroup tuples, add a member, and move a tuple across a frame to change its membership. A group is one framed block with alternating tint, verbalized as one instance at every level of the expression tree, and it travels as one instance through Excel (Group and Ordering columns), .igx, the merge, SBVR individuals, UML object rows and the other exports.
Set-valued columns are picked, not typed — a check list, a literal combo box or a lookup, depending on what the role holds.
The literal in text — a set, bag or sequence can be written as a literal in an expression, in the population dialog, in .exp files and through the MCP tools.
Details page as an entity view (#5991) — the details page shows the selected tuple as a record beside its notes, with the related facts of that entity beneath it, sizeable set grids and data-typed fields. A scenario checklist sits in the editor as well.
The population is checked against the constraints — uniqueness, totality and the well-formedness of tuple groups (kind, head, identity), with counter-examples reported. These population checks are reported as warnings, never as errors, so a work-in-progress population does not block validation.
Fixed: a row disappeared after using the lookup (#5961); the inline combo box was misplaced on scrolled cells.

2. Import an XML Schema as a model

File > Import > XML Schema (XSD)… (#6001) reads an .xsd and builds an information grammar from it: complex types become object types, elements and attributes become fact types, enumerations become value constraints. The importer creates a diagram per schema section and hands off to the Semantic Wizard so you can turn the structural names into proper sentences. Available in every edition.

Also on the import side:

SKOS vocabularies (#5998) — a Turtle file that holds only a SKOS concept scheme (a begrippenlijst) now imports as concepts with their scheme as container, instead of an empty model. Dropped .ttl files take this route automatically.
Recovery instead of refusal — an .igx with dangling references opens in the recovery dialog, where you see and fix what is missing, rather than failing.
Excel import no longer appears frozen (#5819) — progress events are coalesced so the status bar keeps painting during a long import.

3. Delete with your eyes open, check integrity fast

One Delete-from-Repository dialog (#6000) — every Delete, from the Repository Viewer and from the diagram alike, opens the same dialog with a live preview of what the removal will take with it (dependent fact types, constraints, populations, diagram shapes). What the diagram offered before was a subset of what the viewer offered; now they are the same.
Integrity Check (#5999) — shows progress and a busy cursor, and the total-constraint check went from seconds to a fraction of a second on large models.
The 'application seems to be frozen' report during long operations is gone: progress and status events keep the freeze watchdog informed.

4. Teamwork: shared files, better merges, a database wizard

Shared files that follow your login (#5980, Team edition and up) — enterprise and project files (preferences, templates, styles, word lists) are published on the server and synced to every modeler at login or on request. The Manager gained a Shared files page to publish, replace, lock, delete, purge and see the history.
Comments are unbounded on the server — repository comment columns no longer cut long descriptions at 1024 characters.
Merge, last phase: remove what the source no longer has — the IG merge wizard lists the types the source version removed and can remove them from the target as the final step of the merge (#5962). A merge only ever moves a type's namespace version forward; a model carrying an older copy of another model no longer rolls that model back (#5969). Container and template members are included, and one bad substitution no longer discards the import.
Database setup wizard — DbSetup ships with the Team, Server and Beta editions for installing, upgrading and maintaining the SQL Server repository from the layered script package.
Portal on IIS (#6002) — a dedicated installer sets up the Portal as an ISAPI application on IIS, including the application pool and configuration.
The modeler no longer interrupts you at connect time with 'Upgrade the database now?'. The welcome page shows the pending upgrade in the Team column instead.
Uploads from 14.x clients are accepted by the 15.1 database again.

5. Diagrams

One constraint button per kind (#5971) — each button adds the constraint if there is none on the selection and edits it otherwise, and the ring constraint has its own dropdown of the eight ring patterns.
Automatic layout survives a Graphviz warning (#5997) — a warning on stderr used to abort the layout; the result is now used and the warning logged.
ArchiMate view (#5974, #5978) — rules appear as motivation elements, the lines between elements reach the picture, the element name is drawn as a bold title, and an underlined name survives the word-wrap. Layers get their own colours and classes are drawn by concept type.
Modern Diagram (experimental) — Tools > Preferences has a Modern Diagram option that draws the FCO-IM and UML diagrams with a new native engine instead of the legacy drawing and PlantUML. It draws six diagram kinds, ring constraints, collapsed facts, notes and containers, and is held to the legacy renderer condition for condition. It is off by default; try it and report what differs.
Fixed: legend text was painted white on exported PNG and SVG images; a transparent background painted black; the PlantUML package styling did not apply; large fills are fast again.

6. Reports and exports

ORM and FBM, both ways — a NORMA .orm file or an FBM Exchange MetaModel .fbm file opens as a project through File > Open, and File > Export writes both formats. The export was rewritten to carry every constraint, the populations, the identifying facts and the diagrams, with stable identifiers so a round trip does not churn the file. Role value constraints come back as label subtypes.
ArchiMate chapter in the project report — every fact and object type gets an entry (#5977), and the concept type is gated per kind: ELM values on fact and label types, ArchiMate values on concepts and containers, with the layers as nodes in the ontology tree (#5975).
Value constraints read by their kind (#5976) — MIM and SBVR value rules read a range as a range, a key/label domain as a domain, and a state machine as one sentence per transition, instead of a flat list of values.
SBVR rules continue the necessity keyword in lower case.
JSON Table Schema (#5965) is a valid data package again, and the Tables and Columns lists name each column once.
Expressions OTL export (#5970) starts again.
Business Glossary export (#5963) no longer fails on a second language.
Diagrams with embedded pictures no longer exhaust PlantUML's memory (#5790).

7. Expressions and the editor

Third view style for the expression list (#5973) folds the empty Namespace and Version columns away so the sentence gets the width.
A quote or apostrophe in the middle of an expression is text, not a string delimiter (#5972).
A known value is recognised at once while typing a new expression; the 'noname' type that appeared until you pressed Back is gone (#5995).
Derivations as role paths — a derivation rule is recorded as role paths joined on their endpoints (inner, outer, tuple and projection joins), the first step towards derivation rules that generate to SQL and to rule languages.

8. Everyday fixes

Embedded browser pages and the ReSpec project report follow the CaseTalk theme instead of Chromium's guess (#5966, #5968).
Opening a project no longer stops with 'List index out of bounds' while its files are reopened (#5996).
Files below the project folder are stored relative to it, not as shortcuts, and are shown as a virtual subfolder.
The SQL window no longer crashes in idle after a failed query.
A file is judged by what it finally is, in the history and in the project checks.
Markdown editor — find in the rendered page, a filter box above the folder tree, links to files that exist, jump to a line, and a source mode for non-markdown files. The installer registers .md files with MarkdownEdit.
Java (#5960) — CaseTalk ships a jlinked Eclipse Temurin 25 LTS runtime. No Oracle-licensed Java is bundled or downloaded any more.
The About box's Copyrights tab is generated from the software bill of materials, so it lists what actually ships.

Before you upgrade

Repository database upgrade. Team and Server editions need the database moved to version 4.10 (unbounded comments, shared files, collection kinds, 14.x uploads). The welcome page tells you when an upgrade is pending; run it from there or with DbSetup.
Newer file format for tuple groups. Models that use set, bag or sequence roles are written in a newer stream format. Keep every machine that shares such models on 15.1.
Population checks are warnings. If you relied on the population checks turning a model red, note that they now report as warnings.