Skip to content

Releases

v0.6.1

Choose a tag to compare

@github-actions github-actions released this 19 Nov 14:24
586f6e6

Added

  • The docs site draws diagrams in the browser with sirene-core built for WebAssembly: an
    edit button on each sample redraws it as its source changes, and a playground page draws
    Mermaid, PlantUML, or DOT as you type, keeping the source in the page's link.
  • sirene-wasm (wasm/, not on crates.io) exports rendering to JavaScript over the module's
    memory, with no imports; docs/assets/sirene-wasm.js loads it. It replaces wasm-smoke:
    CI renders each file in examples/ with it under Node through that loader.
  • Each release attaches sirene-wasm-X.Y.Z.tar.gz (sirene.wasm and sirene-wasm.js) for
    drawing diagrams in web pages. The loader's renderAll replaces a page's mermaid,
    plantuml, and dot code blocks, and mermaid.js's .mermaid blocks, with their drawings;
    examples/web/index.html is a complete page.
  • The same files are published to npm as @jaemk/sirene, through trusted publishing with
    provenance, so pages can load them from jsDelivr or unpkg. require("@jaemk/sirene") returns
    the loader.
  • Guide pages for the Rust library and for web pages.

v0.6.0

Choose a tag to compare

@github-actions github-actions released this 18 Nov 22:10
03bb390

Added

  • diagram::Lowered::ascii, svg, png (feature png), and scene render a diagram from
    sirene-core alone, with the diagram's own theme applied, as the CLI does. The CLI calls
    them, and a test checks its output against them for each file in examples/.
  • source::read reads a diagram file as UTF-8 of at most 64 MiB, as the CLI and editor do.
  • sirene-core builds and renders on wasm32-unknown-unknown, checked in CI by rendering the
    examples under Node.
  • diagram::Lowered::kind is the type the source was read as, with Kind::lang,
    Kind::name, and Kind::ALL. -v logs it (read as a sequence diagram).
  • The macOS app declares DOT (.dot, .gv) and PlantUML (.puml, .plantuml, .pu,
    .iuml) files, and opens Markdown files as an alternate editor.
  • sirene ascii --width N exits 1 when a drawing is wider than N columns, still printing it.
    ascii::columns measures a drawing's width.
  • CI checks sirene-core's API against its latest release with cargo semver-checks.
  • The editor edits Mermaid sequence diagrams on the drawing: select, rename, add, and delete
    participants and messages, with undo. sequence::edit holds the source edits, and the scene
    owns participants as nodes and messages as edges.
  • The editor edits Mermaid state diagrams on the drawing: rename, add, connect, and delete states
    and transitions, with undo. state::edit holds the source edits.
  • The editor edits Mermaid class diagrams on the drawing: rename, add, connect, and delete
    classes and relations, with undo. class::edit holds the source edits.
  • The editor edits Mermaid ER diagrams on the drawing: rename, add, connect, and delete entities
    and relationships, and set each end's cardinality and whether a relationship is identifying,
    with undo. er::edit holds the source edits.
  • The editor edits Mermaid requirement diagrams on the drawing: rename, add, connect, and delete
    requirements and relationships, and set a relationship's type, with undo.
    requirement::edit holds the source edits.
  • The right-click menu on a class diagram relation holds Kind (association, inheritance,
    composition, aggregation, dependency, realization, link, dashed link).
  • The right-click menu on a class holds Add Attribute, Add Method, and Remove Member.
  • Dragging a state onto another composite state moves it inside.
  • The right-click menu on sequence messages holds an Arrow submenu; Move Earlier and Move Later
    reorder participants.
  • Mindmaps take drawing edits: adding, renaming, and deleting nodes, through mindmap::edit.
  • Timelines take drawing edits: adding, renaming, and deleting periods and events, through
    timeline::edit.
  • Pie charts take drawing edits: adding, renaming, and deleting slices, through pie::edit and
    items::Items.
  • Quadrant charts take drawing edits: adding, renaming, and deleting points, through
    quadrant::edit.
  • User journeys take drawing edits: adding, renaming, and deleting tasks, through
    journey::edit.
  • Gantt charts take drawing edits: adding, renaming, and deleting tasks, through gantt::edit.
  • Kanban boards take drawing edits: adding, renaming, and deleting columns and cards, through
    kanban::edit.
  • Packet diagrams take drawing edits: adding, renaming, and deleting fields, renumbering the
    fields after them, through packet::edit.
  • DOT graphs take drawing edits: adding, renaming, connecting, and deleting nodes and edges,
    through dot::edit.
  • PlantUML use case, component, and deployment diagrams take drawing edits: adding, renaming,
    connecting, and deleting elements and links, through plantuml::edit.
  • sirene ascii --markdown and sirene check --markdown read stdin, or files of any name, as
    Markdown.
  • sirene check --json prints the diagnostics to stdout as one JSON array.

Changed

  • The editor names UTF-16 files and the first invalid UTF-8 byte when it cannot open a file,
    as the CLI does.
  • The editor's file poll checks the file's metadata and reads it only when that moved.

Fixed

  • sirene-core no longer panics laying out a graph or reading a gantt chart on
    wasm32-unknown-unknown, which has no clock.

v0.5.0

Choose a tag to compare

@github-actions github-actions released this 16 Nov 21:37
0145517

Added

  • Logging through SIRENE_LOG or RUST_LOG, with module filters, severity labels, and -vv
    for trace output. Logs go to stderr and keep rendered stdout unchanged.
  • A shared, cloneable sirene_core::Error that retains typed causes and reports the full
    causal chain.
  • Phase timings for reading, layout stages, and drawing, logged at trace level under
    sirene_core::profile (e.g. SIRENE_LOG=off,sirene_core::profile=trace). The stress bench
    prints them per shape with --phases.

Changed

  • Library graph construction and generic ASCII/scene rendering now return results, so invalid
    graph references can be reported to callers. PNG, editor, and release-tool errors retain
    their underlying causes.
  • Production crates reject explicit panic, unwrap, expect, and unreachable calls through
    Clippy; tests can still use them.
  • Faster layout and drawing of large diagrams: edge track assignment orders a gap's routes in
    linear time per relation, sequence lifelines no longer scan every activation per row, ASCII
    canvas cells take half the memory, and long lines of quoted items, xychart categories, or
    ;-joined sequence statements are read in linear time. Output is unchanged.
  • Block, C4, Wardley, journey, git, and styled flowchart drawings look up ids, classes,
    sections, and lanes by index instead of scanning per item; one-line block rows are read in
    linear time; sankey cycle checks skip links that repeat a kept one. Output is unchanged.
  • The editor paints only what lies in the drawing area, and outlines, prunes, zooms to, and
    styles a large selection without looking up each selected node by scanning every node.
  • Diagnostic lines and columns for the editor, check, and Markdown blocks come from one
    diagnostic::LineIndex per source instead of a scan from the start per diagnostic.

Fixed

  • Clicking C4 or architecture nodes no longer indexes an empty flowchart model.
  • Failed terminal writes, directory traversal, confirmation input, and file dialogs are
    reported as errors. Settings persistence failures remain observable in the editor.
  • Parsers reject source beyond their representable byte limit with a diagnostic.
  • Editor: the block added to a Markdown file with none is named for the language of the code
    typed so far (dot, plantuml, mermaid), not the first keystroke's.
  • Desktop app: a writable file in a directory that does not allow a new file is saved in place
    instead of failing.
  • Desktop app: an export to the open document's own file is refused instead of replacing the
    diagram's source with the image.

v0.4.0

Choose a tag to compare

@github-actions github-actions released this 14 Nov 17:52
8a1e987

Added

  • sirene ascii FILE.md prints each diagram block of a Markdown file, with diagnostics at
    positions in the file, instead of reading the whole file as one Mermaid diagram.
  • Mermaid class diagrams: link A "url", click A href "url", callback A "fn", and
    click A call fn(args) set the class's link or callback, with an optional tooltip.
  • Release tarballs are signed with zipsign (cargo xtask sign, cargo xtask gen-key --out DIR).

Changed

  • sirene update checks the release archive's signature against a key built into sirene,
    instead of running gh attestation verify when gh is installed. An unsigned or badly signed
    archive is refused. Releases still carry build provenance attestations.
  • The guide and specs document theme palettes, CLI limits, signing setup and key rotation,
    the pinned toolchain and workspace crates, and sequence numbering while it is off.

Fixed

  • cargo xtask sign uses an exclusively created temporary file, preserving existing files and
    symlink targets beside the archive and retaining the archive's permissions. The release
    workflow decodes its signing key with owner-only permissions.
  • CLI: images, -o files, and --link rewrites are written to an exclusively created temporary
    file and renamed into place, so a failed write keeps the old file and an existing temporary
    file is left alone; a symlinked output is replaced, not written through. Inputs and stdin
    over 64 MiB are refused.
  • The install script stages the binary before replacing it, so a failed copy keeps the old
    binary and a symlink's target is left alone. A directory at the install path is refused.
  • PlantUML YAML: escaped double quotes and doubled single quotes keep #, commas, and colons
    inside the quoted value instead of reading them as comments or separators. An apostrophe in
    plain text (don't) no longer hides a comment or merges flow-array items.
  • Markdown: replacing an unclosed block at the end of a text with empty code no longer panics.
  • PlantUML timing: at most 100 clocks, the rest left out with an error; each state label is
    measured once. Cynefin: many quoted items on one line are read in linear time.
  • Flowchart: at most 100000 edges; a & product past that is left out with one error. Long
    labels, runs of < or &, and many classDef, class and style lines are read in near
    linear time.
  • ASCII output: a wide character written over part of another clears the rest of it, one that
    does not fit is left out, and title and git label widths match the canvas.
  • Block diagrams: space cells past 100000 are left out with one error; unclosed strings and
    classDef lines are read in linear time. Mindmaps: unclosed markdown strings and @{ on many
    lines do not each rescan the rest. Git graphs: branches are looked up by name in constant time,
    and an order too large for a number is an error.
  • C4: boundaries are placed and UpdateRelStyle finds its relations in linear time. ER and
    requirement diagrams: classDef lines are read in linear time. ER: tabs around to and
    optionally to read the same as spaces.
  • SVG output: control characters XML does not allow are written as U+FFFD, and a NaN or infinite
    number as 0, so the file stays well-formed.
  • Layout: rank tightening visits only items next to one that moved and has a work cap, instead
    of sweeping every item until nothing moves; a min_len over 100 counts as 100.
  • Mermaid state diagrams: an unclosed << or [[ ends with its line instead of taking the
    lines up to a later >>.
  • Mermaid sequence, state, and class diagrams: participant, classDef, namespace, and region
    lookups no longer take quadratic time.
  • Desktop app: an export is written to a temporary file and renamed over the target; a file over
    64 MiB is not opened; a settings file that cannot be read is no longer overwritten with
    defaults; a NaN zoom is ignored.
  • PlantUML regex diagrams: a class nested in a class, as in [[:alpha:]], is read as one class,
    and { 2 } with spaces is literal text instead of a repeat, as in Java. Runs of unmatched
    literal { are read in linear time.
  • Mermaid ZenUML: else, catch, and finally may start the line after a branch's }; long
    diagrams no longer lower in quadratic time.
  • Mermaid use case diagrams: a JSON \u surrogate pair is one character, \u takes exactly four
    hex digits, and a JSON table whose rows would hold more than 16 MiB is an error.
  • Mermaid event modeling: a data block line indented with a no-break space keeps its text, and
    the ASCII relation arrows no longer take quadratic time per relation.
  • Mermaid venn diagrams: a union of more than 8 sets, and a size over 1e100, are errors instead
    of a slow layout or NaN geometry.
  • Mermaid Wardley maps: a stage @position outside 0 to 1 or out of order is an error instead
    of overflowing the ASCII drawing; a size outside 100 to 20000 keeps the default; a pipeline
    or evolve declared again keeps its place.
  • Mermaid kanban boards: a long ticket or assignee is cut short with ... instead of
    overlapping the other and the card's edge; in ASCII, a long word is broken to fit the card.
  • Mermaid railroad diagrams: a non-breaking or other Unicode space before an EBNF operator no
    longer panics, and more than 64 operators on one element are an error instead of a stack
    overflow.
  • Mermaid tree views: a huge rowIndent, paddingX, paddingY, or lineThickness is capped
    at 1000 instead of drawing at infinite size.
  • --link: a path with spaces or parentheses is linked as <...>, the link is indented as the
    fence (so it stays in a list item), and a link from an earlier run is replaced when a new block
    shifts the numbering instead of a second link being added. #, ?, %, and : in filenames
    are percent-encoded so the link reaches the image file.
  • Mermaid XY charts: values past the last category no longer stretch the value axis, and on a
    numeric x-axis each series spans the whole range by its own length, as in Mermaid.
  • Mermaid sankey diagrams: a value may have an exponent (1e3); a value over 1e30 is an error
    instead of drawing the chart with NaN geometry.
  • Markdown: diagram blocks inside blockquotes and list items (nested, ordered, or opened on the
    item's line) are found, and an info string may name the language in any case or in braces
    ({.mermaid}, {mermaid}, mermaid{...}). The editor writes a quoted block back with its
    > markers.
  • -o - writes the SVG or PNG to stdout instead of to a file named -.
  • Reading UTF-16 or other non-UTF-8 input fails with a message naming the problem instead of
    stream did not contain valid UTF-8.
  • A missing short file name such as pdf is no longer refused as a mistyped command.
  • Mermaid quadrant charts: x-axis Low --> with no high label draws Low ⟶, as in Mermaid.
  • Mermaid gantt charts: a huge duration is an error instead of overflowing, and excludes is
    not walked over a task longer than 1,000,000 days.
  • Mermaid timelines: a line starting with # is a comment, as in Mermaid, instead of a period.
  • Mermaid radar charts: radar-beta: is a header; options may share a line separated by ,;
    axis and curve ids may hold - and start with a digit; labels may use single quotes and \
    escapes; a title runs past # and ;.
  • Mermaid packet diagrams: labels may use single quotes and \ escapes; a title runs past #
    and ;.
  • Mermaid treemaps: a valueFormat width over 1000 is a warning instead of padding the value
    with that many characters.
  • Mermaid class diagrams: a class belongs to the namespace that declares it with class, even
    if mentioned earlier; relations and A : member lines inside a namespace are warnings and no
    longer pull classes into it.
  • Mermaid sequence diagrams: a box holds only participant/actor declarations, and
    declaring a participant in a second box is an error, as in Mermaid.
  • Mermaid git graphs: commit "message" (without msg:) is read as the message instead of
    dropping the commit.
  • Mermaid user journeys: # comment after a title, section, or task is skipped instead of an
    error.
  • Mermaid pie charts: a slice under 1% is left out and the others fill the circle, as in
    Mermaid, instead of leaving an empty gap.
  • C4 diagrams: UpdateRelStyle finds a Rel_Back by its from and to as written; a
    boundary's { may start the next line, as in Mermaid.
  • Mermaid requirement diagrams: a second definition of a name is a warning and skipped, as
    Mermaid keeps the first, instead of adding its fields to the first.
  • Mermaid architecture diagrams: when two items want the same side of a node, the second is
    placed beside the first instead of beyond it, and an edge turns in the gap between items
    instead of running through one.
  • Editor: a subgraph direction added in a CRLF file keeps CRLF line endings.
  • Mermaid state diagrams: a %% comment may follow a statement or a closing };
    state X:::class (with or without {) sets the class; [[fork]], [[join]] and
    [[choice]] work like <<fork>>, and both forms ignore case. A trailing %% comment is
    also accepted in the other line-based Mermaid diagrams.
  • Mermaid sequence diagrams: non-ASCII text after note left, after participant metadata, or
    in a box title no longer panics; autonumber OFF ignores case.
  • Mermaid ER diagrams: non-ASCII text where a relationship may start no longer panics.
  • Mermaid class diagrams: a member's leading ~ (package visibility) is kept instead of read
    as a generic; a label from class A["Label"] is kept when a later mention has generics.
  • DOT: a repeated edge in a strict graph no longer takes the current edge [...] defaults
    over the first edge's own attributes; \G in a cluster label is the cluster's name;
    style=rounded on a cluster warns as it is not drawn; an edge to a reopened named
    subgraph reaches all its nodes; \\ before a line...
Read more

v0.3.0

Choose a tag to compare

@github-actions github-actions released this 13 Nov 00:10
f72d3a6

Added

  • Graphviz DOT: graph, digraph, and strict sources, .dot and .gv files in
    directories, and dot or graphviz Markdown blocks draw with sirene's layout, including
    clusters, node and edge defaults, record labels as compartments, shapes, colors, arrows,
    and end labels. Attributes that are not drawn are warnings.
  • DOT rank=same subgraphs, constraint=false, and minlen=0: nodes share a rank, and edges
    between them run straight across, below the rank, or up through a label above it.
  • DOT rank=min, max, source, and sink subgraphs: nodes go in the first or last rank,
    or a rank of their own above or below every other node.
  • DOT compound=true with lhead and ltail: edges end at a cluster's border.
  • DOT ordering=out and ordering=in on the graph or a node keep its edges in source order.
  • DOT node group: edges between nodes of one group run straight.
  • DOT compass ports (a:e, a:p:sw, tailport, headport) move an edge end toward that end
    of the node side.
  • DOT record ports (a:p, <p> fields) join the field's line in rankdir=LR and RL graphs.
  • DOT HTML table node labels draw as compartments, with <HR/> sections and PORT cells.
  • --lang mermaid|dot|plantuml reads every diagram as that language. Without it, a file's
    extension (.dot, .gv, .mmd, .mermaid, .puml, .plantuml, .pu, .iuml) or a
    Markdown block's name picks the language in the CLI and the editor, and other sources are
    detected from their first line.
  • PlantUML sequence diagrams (@startuml), drawn by the sequence layout and read-only in the
    editor: participant kinds, aliases, order, boxes, arrows with crosses, edges ([->, ->]),
    ++/--/**/!!, return, autonumber, notes, frames, group, separators (== x ==),
    delays (...), spaces (|||), ref over, and hide footbox. Constructs that are not drawn
    are warnings, and other PlantUML diagram kinds are errors.
  • PlantUML activity diagrams, drawn by the layered layout: actions and their shapes, start,
    stop, end, if/elseif/else, switch, while, repeat with backward, break,
    fork, split, partitions and groups, swimlanes (|Lane|), arrow labels and styles, and
    notes. The diagram kind is detected from the first statement as PlantUML does.
  • PlantUML class and object diagrams, drawn with compartments: class kinds and stereotypes,
    members with {static}/{abstract} and separators, extends/implements, links with every
    head, multiplicities, direction words, and -[#color,dashed]-> styles, packages, objects,
    maps, hide empty members, and notes.
  • PlantUML state diagrams: start and end states, descriptions below the name, colors,
    <<choice>>, <<fork>>, <<join>>, history states, composite states with concurrent
    regions (--, ||), styled transitions, and notes.
  • PlantUML use case, component, and deployment diagrams: actors, use cases, components,
    interfaces, and the deployment element shapes (node, database, cloud, queue,
    artifact, ...), the :x:, (x), [x], and () x short forms, stereotypes, colors,
    descriptions, nested containers, styled links, notes, and skinparam actorStyle.
  • PlantUML mindmaps (@startmindmap) and work breakdown structures (@startwbs): *, +,
    and - node lines or indented * lines, boxless nodes, multi-line :text;, WBS colors,
    aliases, and (A) -> (B) arrows.
  • PlantUML Gantt charts (@startgantt): tasks with durations, start and end dates, then,
    references to other tasks' starts and ends, [A] -> [B], milestones, completion, sections,
    closed weekdays and dates, printscale, and today is.
  • PlantUML JSON and YAML (@startjson, @startyaml), drawn left to right as a table per object
    or array with lines to nested tables.
  • Compartment nodes with an empty label have no title area.
  • PlantUML @startdot (a Graphviz graph), @startebnf (drawn as a railroad diagram), and
    @startfiles (paths drawn as a tree view, with <note> descriptions).
  • PlantUML @startregex: each line's regex drawn as a railroad track, with classes, escapes,
    groups, choices, and quantifiers. Railroad rules with no name draw no name =.
  • PlantUML @startchen: entity-relationship diagrams in Chen notation, with weak entities,
    identifying relationships, key, derived, multi-valued, and composite attributes,
    cardinalities, total participation, and specializations.
  • PlantUML network diagrams (nwdiag): networks with addresses, hosts with shapes and
    descriptions, groups, and peer links.
  • PlantUML @startchronology: [Event] happens on DATE events drawn as a timeline in date
    order.
  • PlantUML timing diagrams: robust, concise, rectangle, binary, clock, and analog signals,
    absolute, relative, named, clock-based, and date moments, hidden and undefined states,
    messages, constraints, highlights, and notes, drawn on a time axis in SVG, PNG, and text.
  • Notes on composite states in state diagrams, and class diagram notes with no class
    (note "text"), which draw as a note box of their own.
  • Mermaid themes: a diagram's theme and themeVariables (including fontSize) from its
    frontmatter config: or an %%{init}%% directive set its colors in SVG, PNG, and the editor,
    for all 11 Mermaid 12.3 themes; an unknown theme name is a warning. Pie slices and the
    sections of mindmaps, timelines, radar charts, and treemaps take the theme's pie and
    cScale colors.
  • sirene update replaces the running binary with the latest release (--check, --to VERSION, --yes), checked against SHA256SUMS and its build provenance; it refuses for a
    Homebrew or cargo install binary and names the command to use instead.
  • Mermaid state diagrams (stateDiagram, stateDiagram-v2) in every command and in the
    editor, where the drawing is read-only: states, transitions, [*], composite states,
    concurrent regions, fork, join, and choice, notes, direction, and styling.
  • Mermaid class diagrams (classDiagram, classDiagram-v2), read-only in the editor: classes
    with attribute and method compartments, annotations, generics, every relation with its
    markers, cardinalities, and label, namespaces, notes, and styling.
  • Mermaid ER diagrams (erDiagram), read-only in the editor: entities with attribute columns
    (type, name, keys, comment), aliases, relationships with crow's-foot markers in symbol or word
    form, and styling.
  • Mermaid git graphs (gitGraph), read-only in the editor: commits, branches, checkout, merges,
    and cherry-picks on lanes in each direction (LR, TB, BT), with commit types and tags; in
    text, one commit per row as git log --graph draws them.
  • Mermaid timelines (timeline, LR and TD), read-only in the editor: periods with events,
    continuation lines, sections with their own colors, <br>, and wrapping, in images and text.
  • Mermaid sequence diagrams (sequenceDiagram), read-only in the editor: participants, actors,
    and the typed shapes, every message arrow, self-messages, activations, notes, loop, alt,
    opt, par, critical, break, and rect blocks, box groups, create and destroy,
    and autonumber, in images and text.
  • Mermaid pie charts (pie), read-only in the editor: slices with percentages, a legend,
    showData, and a title; in text, a table with a bar per slice.
  • Mermaid gantt charts (gantt), read-only in the editor: tasks with after, until,
    durations, and tags (active, done, crit, milestone, vert), sections, excludes,
    includes, weekend, inclusiveEndDates, todayMarker, topAxis, compact rows, and any
    dayjs dateFormat and d3 axisFormat, also settable in frontmatter or a directive.
  • Mermaid mindmaps (mindmap), read-only in the editor: a tree built from indentation, every
    node shape, markdown strings, and ::: classes, drawn out to both sides of the root with a
    color per branch; in text, an outline of the tree.
  • Mermaid user journeys (journey), read-only in the editor: tasks in sections with a score
    from 1 to 5 drawn as a face, and actors as colored dots with a legend; in text, a score bar per
    task.
  • Mermaid quadrant charts (quadrantChart), read-only in the editor: axis and quadrant labels,
    and points styled directly or by classDef (radius, color, stroke-color,
    stroke-width); in text, the points plotted by name in a box split in four.
  • Mermaid XY charts (xychart-beta, xychart), read-only in the editor: bar and line series
    over category or numeric x-axes, vertical or horizontal, with axis titles and ranges, point
    labels, and a legend for named series; in text, a row per position with bars and markers.
  • Mermaid requirement diagrams (requirementDiagram), read-only in the editor: every
    requirement type and elements as boxes with their fields, the seven relationship types in
    either direction, classes, styles, and direction.
  • Mermaid kanban boards (kanban), read-only in the editor: columns of cards with assigned,
    ticket, and priority metadata, drawn with a priority stripe; in text, a box per column.
  • Mindmaps read @{ ... } node metadata and warn that it is not used.
  • Mermaid packet diagrams (packet, packet-beta), read-only in the editor: bit ranges, single
    bits, and +count fields in rows of bitsPerRow, with bit numbers and bitOrder: descending;
    in text, the RFC header layout.
  • Mermaid radar charts (radar-beta), read-only in the editor: axes, curves with listed or
    keyed values, min, max, ticks, circle or polygon graticule, a legend, and the
    frontmatter radar config; in text, a bar per axis and curve.
  • Mermaid sankey diagrams (sankey-beta, sankey), read-only in the editor: CSV rows of flows
    laid out as d3-sankey does, with linkColor, nodeAlignment, showValues, prefix, and
    suffix; in text, each node's outgoing flows with bars.
  • Mermaid treemaps (treemap-beta, `tree...
Read more

v0.2.0

Choose a tag to compare

@github-actions github-actions released this 09 Nov 03:38
c9d81dc

Added

  • A documentation site at https://jaemk.github.io/sirene/: a landing page and an mdBook
    guide, with the examples rendered by sirene at build time.
  • sirene png --quality draft|normal|fine (default fine, 4x); the default scale shrinks
    to fit the largest image.
  • sirene svg DIR and sirene png DIR render every .mmd and .mermaid file under a
    directory, beside each or under -o OUTDIR.
  • -w/--watch on ascii, svg, and png renders again when an input changes.
  • -v/--verbose logs each step, with timings, to stderr.
  • Markdown bold and italic in labels; long labels wrap.
  • examples/ with diagrams to try.
  • Editor: fit large drawings on open, zoom to selection, arrow-key panning, select all, and
    tab/shift-tab through nodes.
  • Editor: File > Export as PNG or SVG, and Edit > Copy Image.
  • Editor: View > Appearance (Follow System, Light, Dark), File > Open Recent, and a resizable
    source pane, kept in a settings file with the window size.
  • Editor: the menus are drawn in each window on Linux and Windows, where there is no system
    menu bar.
  • Editor: Edit > Cut, Copy, and Paste; Undo, Redo, and Select All act on the source text when
    it has the keyboard.
  • Editor: a focus ring shows when the drawing has the keyboard.
  • Editor: a right-click menu on the drawing to set node shapes, edge lines, arrowheads, and
    lengths, reverse edges, add a node before or after, and connect to or from any node.
  • flowchart::edit::set_node_shape, set_edge_line, and reverse_edge.
  • Editor: a shape bar on the drawing that adds a node of a shape; double-click empty space to
    add a node, or drag from a node to empty space to add a linked one.
  • Editor: layout hints: set the diagram's or a subgraph's direction, group and ungroup nodes,
    move nodes into or out of subgraphs (also by dragging onto one), keep two nodes together, and
    move statements earlier or later.
  • flowchart::edit::set_direction, keep_together, group, ungroup, move_to_subgraph,
    move_node, and move_edge. Ungroup sends edges at the subgraph to its members, and moving
    a node out of a subgraph takes its chains with it.
  • Editor: a disabled right-click item says why it is disabled, and a refused key shows the reason
    in the status bar.
  • Editor: a styles panel (View > Show or Hide Styles) that applies classes to the selected
    nodes and edits a class's colors, width, and dashes.
  • flowchart::edit::set_class_style, apply_class, remove_class, and fresh_class_name.
  • Editor: the styles panel sets the selected edges' color, label color, width, and dashes with
    linkStyle; flowchart::edit::set_link_style.
  • sirene check FILE... prints diagnostics only, for diagrams, directories, and Markdown files.
  • sirene svg FILE.md and sirene png FILE.md write one image per mermaid block; --link
    adds an image link after each block, keeping the blocks.
  • sirene_core::markdown::mermaid_blocks and flowchart::Parse::other_diagram.
  • Editor: sirene FILE.md edits a Markdown file's mermaid blocks one at a time, writing each
    edit back into its block; markdown::Block::replace_code.
  • The repo is a mind source (mind meld jaemk/sirene) offering the sirene:render skill, which
    teaches agents to check and render diagrams with the CLI.

Changed

  • sirene png writes at 4x by default (was 2x); --quality normal gives the old size.
  • The source pane starts 480px wide and keeps its width when the window resizes.
  • New windows open offset from the last one instead of on top of it.
  • Other Mermaid diagram types (sequenceDiagram, classDiagram, ...) are one
    "not supported" error and draw nothing, instead of drawing their text as flowchart nodes.
  • A mistyped subcommand (sirene asci FILE) is an error with a suggestion instead of opening
    windows.
  • Save As and Export start in the home directory when the app runs from /.

Fixed

  • Readable label colors on styled fills; straighter edges between near-aligned nodes.
  • svg and png refuse to overwrite their input.
  • New windows fit the screen when the remembered size came from a larger one, and a remembered
    source pane leaves the drawing at least 240px.
  • Settings are saved through a temporary file, so a crash cannot leave them half written.
  • svg DIR and png DIR skip a diagram whose output another already writes (a.mmd beside
    a.mermaid) instead of overwriting it, and report an unreadable DIR or an -o that is a file
    once instead of per diagram.
  • Saving keeps file permissions, owner, group, and symlinks; a read-only file is not saved.
  • Reload asks before dropping unsaved edits; Save asks before replacing changes made on disk.
  • Opening a file that is already open brings its window forward.
  • Deleting a node drops the id@{ ... } and class statements of its edges.
  • The label field covers the label it edits, with its text centered, instead of showing it
    through a translucent background in the dark theme.

v0.1.1

Choose a tag to compare

@github-actions github-actions released this 08 Nov 02:35

Added

  • Release binaries for macOS (aarch64, x86_64) and Linux (aarch64, x86_64; gnu
    with the GUI, musl without), with SHA256SUMS and build provenance
    attestations.
  • bin/install.sh, a one-line installer for the release binaries.
  • Homebrew: this repository is a tap, with a sirene formula for the binaries
    and a sirene cask for Sirene.app.

Changed

  • The macOS bundle identifier is io.github.jaemk.sirene, and the Mermaid
    document type is io.github.jaemk.sirene.mermaid.

v0.1.0

Choose a tag to compare

@github-actions github-actions released this 08 Nov 01:10

sirene-gui-v0.1.0

Choose a tag to compare

@github-actions github-actions released this 08 Nov 01:10

sirene-core-v0.1.0

Choose a tag to compare

@github-actions github-actions released this 08 Nov 01:10