interpres/CHANGELOG.md

Changelog

All notable changes to interpres are documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[development]

Added

• encoding.TextMarshaler and encoding.TextUnmarshaler are honoured by default, with no option to switch them off. A type that implements them is encoded as a TOML string and decoded from one: net.IP becomes "192.0.2.1", and a user type with MarshalText or UnmarshalText follows. MarshalTOML and UnmarshalTOML still win over the text methods, and the four date-time types keep their bare timestamp form instead of becoming a quoted string. A struct type that implements the interface now encodes as a string where it was a table before, which is the breaking part of the change.
• time.Duration is encoded in its canonical Go form as a TOML string, 1h30m0s, because TOML has no duration type; the decoder reads that string back and still accepts a bare integer as the nanosecond count.
• interpres-decode -encode, the adapter's other direction: it reads the toml-test tagged JSON from stdin and writes the TOML document it describes. The compliance suite now runs the encoder as well as the decoder, 214 encoder cases against the tagged JSON of the valid corpus.
• InlineTables(threshold): a sub-table whose single-line rendering is at most threshold bytes is written as an inline table instead of a header section, which shortens a document of small tables. An array of tables keeps its header form, because its inline form would re-parse as a value array.
• Document and ParseMap: Parse now returns a *Document, which holds the values together with the key order, whether a table was written as an inline table or under a header, and the comments, with Keys, Entries, Get, Comments and SetComments to read and write them. ParseMap returns the plain map[string]any tree, the shape Parse used to give.
• The edit pipeline on a Document: typed getters on Document and Table (GetString, GetInt, GetFloat, GetBool, GetArray, GetTable), Set and Delete that keep the surviving keys' positions and comments, UnmarshalDocument, which decodes the document into a typed destination without parsing again, and Marshal of a Document, which writes the keys in written order, the comments above the lines and headers they belonged to, and the inline tables inline.
• OffsetDateTime, the Go type of the offset date-time kind, so that all four TOML date-time kinds have one of their own. Parse and Unmarshal hand it back where they produced a bare time.Time before, and Marshal accepts it. Unmarshalling into a struct field of type time.Time keeps working, because the plain type takes an offset date-time as it always did; code that asserts the tree's type, and UnmarshalTOML implementations that expect a time.Time, need the new type.
• MaxNestingDepth(depth) and MaxInputSize(size) options bound the parse a Decode performs, and every parse carries a nesting limit in any case (10000 levels, which no hand-written document approaches): a document that nests arrays or inline tables deeper used to run the stack out and is now rejected with a SyntaxError naming the limit.
• Statements(r), an iterator over the top-level statements of the document the reader carries, in written order: key/value statements, a [table] header as one statement with its node, an [[array of tables]] as one statement per element. A caller that breaks after the statement it wanted reads no further ones. examples/statements shows the walk.
• ParseAs[T](data), the generic one-line decode, and NewSchema[T](), which precompiles the struct schema and the interface flags for a hot path before the first document arrives.
• EmitFieldComments(true) prints the comment a field's toml tag carries in a comment= option above the field's line or header, the comments a round trip through the Go type would otherwise drop. Go doc comments are not visible to reflection, so the tag is the channel that carries the text.
• LocalTimeLocation(loc) lets a local date-time fill a plain time.Time destination in the location given, relabelled rather than shifted: 07:32 in the document is 07:32 in the zone. Without the option the wrapper types remain the only destinations a local kind fills.
• The parse checks its context inside a value as well as between statements: an array, an inline table and a multi-line string check every 64 elements or lines, so one huge value cannot hold the parse past its cancellation.
• OrderedMap, the string-keyed table that remembers the order its keys were set in: decoding into one fills it in the order the document wrote the keys, and Marshal writes one back in that order, where a map carries no order on decode and sorts on encode. It works as a decode target on its own, in a struct field, and as the element of an array of tables; its values are untyped, so a nested table stays a map[string]any.
• Unmarshal(data, v, opts...) and the other entries take variadic options, the shape encoding/json/v2 uses: RejectUnknownFields, NumbersAsLiterals, MaxNestingDepth, MaxInputSize, LocalTimeLocation. MarshalWrite(w, v, opts...) and UnmarshalRead(r, v, opts...) are the streaming forms.
• Marshal carries a nesting limit of 10000 levels, the parser's own figure: cyclic data, which used to run the stack out, is now rejected with an error that names the limit and the path it was met at.
• The toml tag gained the required option: a field tagged toml:"host,required" makes the decode fail with missing required key "host" when the document carries no key that resolves to it. The option shapes decoding only, and the encoder ignores it.
• UnmarshalerContext, the custom-decode interface that hands the decode's context to the method, UnmarshalTOMLContext(ctx, data). It wins over UnmarshalTOML when a type implements both, so a long custom decode can abort on cancellation; a non-cancellable entry point hands in context.Background, never nil.
• A TOML array decodes into a Go fixed-size array, [N]T, where only a slice was accepted before; the encoder could already encode one. A length mismatch is an error wrapped with the key path.
• MarshalAppend(buf, v) appends the TOML encoding of v to buf and returns the extended buffer, the shape json.MarshalAppend has.
• interpres-decode -version prints the binary's version, the module version the toolchain recorded, so a release-built binary names its own tag.
• interpres-decode -json prints plain indented JSON instead of the tagged form, the shape for people and diffs, with the date-time wrappers in their TOML form.
• interpres-decode -validate walks a named directory for .toml files and closes the sweep with a summary naming how many documents were checked and how many were invalid; single files stay quiet on success as before.
• interpres-decode -struct infers a Go struct definition from a document: one field per key in written order, nested tables as nested struct types, an array of tables as a slice. The printed type compiles and decodes the document it came from.
• interpres-decode -schema TYPE file.go writes a TOML template for the named struct type of a Go source, the comment= tag option printed as a comment and the default= option as the value. It is the inverse of -struct.
• ParseFile(path) reads the file and parses it into a Document, with the file name at the front of every error it returns, read failure and parse failure alike. Valid(data) reports whether a document parses, nil on success and the parse error on failure, the library call the -validate mode of interpres-decode is built on.
• SyntaxError carries the byte Offset the scan stopped at and the 1-based Column on the line, beside the line it always had, and SourceLine(src) renders that line with a caret under the position, for messages shown under the input. An input that is not valid UTF-8 names the offset of the first invalid byte in its message. The new fields are additive: a SyntaxError built from a line and a message alone is unchanged.
• NumbersAsLiterals(true) decodes the integers and floats of the document into Number, which carries the literal the document wrote, so 0x1f, 1_000, +1.0 and inf survive a round trip with their spelling instead of the normalised 31, 1000 and 1.0. Typed destinations take the evaluated value as before, a Number field takes the literal, and Marshal writes a Number back as its bare literal, rejecting one that is not a valid TOML number.

Changed

• The stateful Decoder and Encoder of 1.x are replaced by variadic options on the entries, the shape encoding/json/v2 uses: Layout(kind) with LayoutKindGrouped or LayoutKindDeclaration, OmitEmptyArrays, LiteralMultiline(threshold), InlineTables(threshold), EmitFieldComments, RejectUnknownFields, NumbersAsLiterals, MaxNestingDepth, MaxInputSize, LocalTimeLocation.
• DecodeError and EncodeError carry one Path type, a list of segments ("items", "[0]", "weight") with a String() rendering the TOML notation, items[0].weight. The decode error used to hold a bare []string, the encode error a plain string. Both messages render the same way now, interpres: items[0].weight: ..., with one interpres: prefix where the composition used to double it.
• omitempty follows the encoding/json semantics: the field is skipped when it holds an empty string, a zero number, false, a nil pointer or interface, or a nil or empty slice, array or map. In 1.x the option covered only the collections.
• The toml tag gained the inline option: a struct or map field tagged toml:"retry,inline" emits as retry = {…} instead of a header section, whatever its size, a named embedded struct included. Forcing it on an array of tables is an error, because the inline form would re-parse as a value array and change the value's Go type.
• The output takes the TOML 1.1 form. A date-time writes its seconds only when the value carries them and drops the trailing zeros of a fractional second, so 07:32:00 is written 07:32 and half a second as 00.5. Both are the same value, and a document written without seconds now comes back without them. LocalDateTime.String(), LocalTime.String() and the offset date-time rendering follow the same rule.
• An inline table that would pass the hundredth column is written across lines with a trailing comma and one tab of indentation per nesting level, the shape TOML 1.1 allows an inline table to take.
• MarshalTOML reaches every array element and every field, whatever the Go kind, and its result is normalised like any other value: an element rendering itself as a table keeps the [[header]] form, one rendering itself as a scalar turns the array into a value array, and the method runs once per element. It is found on the addressable pointer as well, so a pointer-receiver MarshalTOML is called for a field or an element, exactly as MarshalText is.
• TOML 1.1 is the acceptance contract, and TOML 1.0 is not. The compliance suite runs the 1.1 corpus alone, and the promise that every 1.0 document parses exactly as before is withdrawn. Nothing that parses today stops parsing: the 1.0 valid corpus still passes in full. The documents whose verdict changes are the ones 1.1 relaxed, such as the \xHH escape sequences 1.0 rejected.
• The module path carries the /v2 suffix the Go toolchain requires of every major version 2 module: imports change to sourcedock.dev/petrbalvin/interpres/v2.
• Input that is not valid UTF-8 is now rejected where the parser's scan meets the invalid byte, with a SyntaxError naming that line, instead of a whole-input check that always reported line 1. Invalid input is still rejected; the reported location is now the byte's own.

Performance

• Struct destinations decode directly: for a type the direct skeleton can model, the parser resolves tables and keys against the struct schema while the document scans and no intermediate value tree is kept. The strict decode of the representative document drops from 168 to 160 allocations per call against the tree path in the same process, and the 2000-element document reaches allocation parity; every document the skeleton cannot model falls back to the tree path and its exact error contracts. A differential fuzz target decodes every generated document both ways.
• Marshal writes plain scalars and typed scalar arrays straight from their reflect cells instead of boxing them into interface values first, and skips the per-element resolution for arrays that can never take the [[header]] form. The representative document now costs 130 allocations per call instead of 141, the long array-of-tables document 55 915 instead of 63 660, with byte-identical output.
• Parsing is faster than in 1.1.0 while carrying the new document layer: the suite's representative document decodes at about 79 MB/s with 104 allocations per call, and the long array-of-tables document at about 106 MB/s against 56 MB/s in 1.1.0, with allocations on that document halved from 67 664 to 31 765. Date-time tokens are validated by a byte scan instead of regular expressions, repeated keys share one string across array-of-tables elements, and per-statement buffers are reused.
• Typed decoding is 12 percent faster than in 1.1.0 on the representative document (9792 ns against 11 147 ns) with 24 percent fewer allocations (167 against 220); interface lookups resolve through a cached per-type flag set instead of boxing every value into an interface to ask.
• Marshal runs at the 1.1.0 speed while emitting the new TOML 1.1 output form, at half the bytes per operation (6170 against 11 348 on the representative document), and writes through a pooled output buffer with a 1 MiB retention cap; repeated marshals keep the live heap flat.
• Two benchmarks measure the shapes that drove the work: BenchmarkStrictDecodeLong and BenchmarkMarshalLong run the 2000-entry document at about 3.8 ms and 3.4 ms per call, at 63 772 and 63 660 allocations.

Fixed

• An offset date-time written with the +00:00 offset kept the anonymous location time.Parse invents for it, so a round trip through the tree and Marshal, which writes a zero offset as Z, changed the value's reflection-visible location. The zero offset normalises to time.UTC at parse, and the tree is stable across the round trip.
• Decoding into a defined type whose underlying kind is string or bool, such as type Name string, panicked instead of storing the value, because a value of the predeclared type is not assignable to a defined type and the decoder assigned it without a conversion.
• A top-level value the encoder could not normalise reported its path with a leading dot, interpres: .port: ...; the message now reads interpres: port: ..., the shape EncodeError.Path already used.

Migration from 1.x

The module path. 2.0 lives at sourcedock.dev/petrbalvin/interpres/v2, the suffix the Go toolchain requires of every major version 2 module. Change every import and go get line:

go get sourcedock.dev/petrbalvin/interpres/v2

TOML 1.1 only. The acceptance contract is the TOML 1.1 corpus, and the promise that every 1.0 document parses exactly as before is withdrawn. Documents whose verdict changes are the ones 1.1 relaxed: \e and \xHH escapes, times without seconds, multi-line inline tables with comments and a trailing comma. Nothing that parsed in 1.x stops parsing, because the 1.1 grammar contains the 1.0 one.

The output takes the 1.1 form. A date-time writes seconds only when the value carries them, a fraction drops its trailing zeros, and a long inline table breaks across lines. A document written from the same value can come out shorter; it re-parses to the same value.

Text methods on by default. A type implementing encoding.TextMarshaler or encoding.TextUnmarshaler now takes the text path with no option to switch it off. A struct that implemented the interface encodes as a string where it was a table before. MarshalTOML and UnmarshalTOML still win.

One Go type per date-time kind. Offset date-times hand back OffsetDateTime, not a bare time.Time. Code that type-asserts the tree or expects time.Time inside UnmarshalTOML needs the new wrapper; a destination field of type time.Time keeps working.

The document carries what the map could not. Parse returns a *Document with the key order, the inline distinction and the comments; ParseMap gives the plain map[string]any tree the old Parse returned. The document is writable, and Marshal writes it back with its comments.

Options instead of Decoder and Encoder. The stateful types of 1.x are gone; the entries take variadic options, the shape encoding/json/v2 uses. NewDecoder().DisallowUnknownFields().Decode(data, &cfg) becomes Unmarshal(data, &cfg, RejectUnknownFields(true)), and the encoder methods become options: Layout(LayoutKindDeclaration) replaces GroupByKind(false), LiteralMultiline replaces UseLiteralMultiline.

Tag options. omitempty follows encoding/json: it now also drops empty strings, zero numbers, false, nil pointers and nil interfaces. required demands a key at decode. inline forces the inline table form at encode. comment=text carries a comment EmitFieldComments prints.

Errors. DecodeError.Path is a Path (segments with a String() renderer), EncodeError.Path the same type instead of a plain string, and both messages render interpres: server.ports[2]: ... with one prefix. SyntaxError gained Offset, Column and SourceLine. Decode errors into a Number-carrying tree and the fixed-size array decode are new shapes a match on the old messages would not see.

Decoding shapes. map[string]any values merge into a non-empty map destination; untagged embedded maps beyond the first stay empty; numbers can stay literals under NumbersAsLiterals; local date-times can decode into time.Time under LocalTimeLocation. All three are opt-in or additive except where noted above.

[1.1.0] - 2026-09-18

Added

• TOML 1.1 support, on by default: date-times and times without seconds (07:32, 1979-05-27T07:32, normalised to full seconds on output), the \e and \xHH escape sequences, and multi-line inline tables with comments and trailing commas. The compliance suite runs in TOML 1.1 mode: 214 valid and 467 invalid cases, zero failures. Every TOML 1.0 document parses exactly as before.
• interpres-decode -validate [file ...]: a validate mode beside the toml-test adapter. It parses each named file, or stdin when none are named, prints one line per invalid document to stderr, and exits 0 when all are valid, 1 when one is not, and 2 on a usage or read failure. Install it with go install .../cmd/interpres-decode@latest; releases still ship no binaries.
• DecodeError and EncodeError: decode and encode failures are wrapped in typed errors carrying the key path, read with errors.AsType instead of parsing the message text. The rendered messages keep their shape; the only visible change is that an encode failure on a top-level field no longer gains a meaningless leading dot in its path.
• omitzero and omitempty tag options on encode: toml:"name,omitzero" skips a field whose value is the zero value of its type (a type with an IsZero() bool method decides through the method), and toml:"name,omitempty" skips a nil or empty slice, array, or map. The decoder ignores both options.

Changed

• The compliance suite is toml-test v2.2.0, up from v1.6.0. Its TOML 1.0 corpus holds 205 valid and 474 invalid cases (185 and 371 before), and it caught the two documents the parser still accepted, fixed below.
• The flattened struct layout the decoder consults is cached per struct type and shared with the encoder, which now resolves duplicate field keys with it. Strict decoding of an array of tables of structs runs about a quarter faster; marshalling structs gained the same layout without measurable cost.
• The parser scans the input bytes in place instead of building a []rune copy of the document: every character that drives the grammar is ASCII and the input is validated UTF-8 up front, so the conversion pass and its four bytes per rune were pure overhead. Parsing a large array-of-tables document runs about a fifth faster and allocates about half the memory.
• Numeric tokens without underscores skip the normalising rebuild: digits are validated in place in joinDigits, and a float whose token is already clean goes to strconv.ParseFloat directly. One allocation per integer atom and two per float atom disappear.

Fixed

• A MarshalTOML result of nil with a nil error fails the marshal with MarshalTOML returned a nil value. The field silently vanished before, and inside a value array the nil result reached reflection as a zero value and panicked.
• Strict decoding reports the smallest unknown key. Several unknown keys in one table made the message depend on Go's random map iteration order, so the same document reported different keys across runs.
• Decoding into a struct that embeds a pointer to itself terminates. The schema walk recursed through the embedded type forever, so such a Unmarshal call hung the process; the walk now tracks the struct types on the current path and stops when one repeats.
• An array-of-tables header whose path runs through an inline table (a = {b = {}} followed by [[a.b.c]]) is rejected. The frozen-inline-table check covered [table] headers and dotted keys but not the intermediate steps of an array-of-tables header, so such a document silently extended the inline table.
• A new element of an array of tables starts a fresh scope for dotted-key paths and nested arrays of tables: [[a]], b.c = 1, [[a]], [a.b] parses, as the TOML examples in the spec shape it. The records of the previous element falsely rejected the same paths in the next one.
• Marshal emits exactly one key when two struct fields resolve to the same TOML name, picking the field the decoder would fill (the shallower one, the later declaration at equal depth). Such a struct previously marshalled into a duplicate key, and the output never re-parsed, breaking the round-trip guarantee.
• Marshal returns an error for a table header key or an inline-table key that is not valid UTF-8, the way scalar keys already did, instead of silently emitting corrupt TOML (a header that lost its key, an inline table with a missing key).
• UseLiteralMultiline falls back to the escaped basic string when the value cannot be carried verbatim by the literal form: a run of three single quotes, a control character, or a lone carriage return. Such values previously produced output that did not re-parse.
• A []any holding only tables marshals in the value-array form with inline tables, keeping the type Parse produces for such an array. It previously took the [[header]] form, so a round-trip changed the value's type from []any to []map[string]any.
• Decoding into a uint destination checks the type's platform width instead of only the fixed widths, so a 32-bit uint no longer truncates silently; decoding a finite float beyond the float32 range is an overflow error instead of a silent infinity.
• Struct fields that resolve to one key at equal depth decode through the field declared later, matching the documented rule; the first one won before.
• A float with an exponent marker but no digits (1e, 0.0E) is rejected; the exponent requires at least one digit.
• A date-time offset outside 00:00 through 23:59 is rejected; such offsets were accepted and silently rolled over (+00:60 decoded as +01:00).
• Untagged embedded fields now decode symmetrically with encode: an embedded struct receives its keys inline (a nil embedded pointer struct is allocated), an embedded map catches the keys no field claims, and a name clash resolves in favour of the shallower field. A struct with an untagged embedded field previously decoded with all inline keys dropped and did not round-trip.
• Marshal re-emits arrays that mix tables with scalars: the table elements render as inline tables inside the value array. A tree that Parse accepts from such a document previously failed with cannot encode map[string]interface {}.

[1.0.0] - 2026-08-20

First stable release: a dependency-free TOML 1.0 parser and encoder for Go that uses only the standard library and passes the entire toml-test suite (185 valid and 371 invalid cases, 0 failures).

Added

Parsing

• Parse(data []byte) (map[string]any, error): decode a TOML document into an untyped tree.
• Full TOML 1.0 syntax: comments; bare, quoted, and dotted keys; tables ([a.b]) and arrays of tables ([[a]]); basic and literal strings, including multiline (""" / ''') with escape sequences and line-ending backslash trimming; integers in decimal, hex (0x), octal (0o), and binary (0b) with _ separators; floats with exponents and inf / nan; booleans; offset/local date-times, dates, and times; arrays; and inline tables.
• Distinct date-time types: offset date-times decode to time.Time, while LocalDateTime, LocalDate, and LocalTime represent the local variants, each with a String() method returning the TOML-canonical rendering.
• Strict, spec-conformant validation that rejects invalid documents: leading zeros, misplaced underscores, malformed floats and radix literals, control characters in strings and comments, bare carriage returns, non-UTF-8 input, out-of-range Unicode escapes, multiline strings used as keys, single-digit date-time components, duplicate/overwriting inline-table keys, and the full family of table redefinitions (header vs. header, header vs. dotted key, array of tables vs. table, and dotted-key appends to defined tables).
• SyntaxError carrying the 1-based line of a malformed document.

Decoding

• Unmarshal(data []byte, v any): parse and map onto a struct or map[string]any via reflection, with overflow-checked numeric conversion, nested structs, slices, and map[string]T.
• toml:"name" field tags, case-insensitive name fallback, and toml:"-" to skip a field.
• RejectUnknownFields(true) option for strict decoding that rejects keys without a destination field, at every struct depth.
• Unmarshaler interface (UnmarshalTOML(data any) error) for types that take full control of their decode.

Encoding

• Marshal(v any) ([]byte, error): encode a struct or map[string]V value to a TOML 1.0 document that re-parses to an equivalent value tree.
• Encoder with chainable policy options: GroupByKind (group-by-kind layout versus declaration order), OmitEmptyArrays, and UseLiteralMultiline.
• Marshaler interface (MarshalTOML() (any, error)) for types that need a custom TOML shape; the returned value is encoded normally.

Cancellation

• ParseContext, UnmarshalContext, MarshalContext, (*Decoder).DecodeContext and (*Encoder).MarshalContext honour a context.Context, checked up front and every 64 statements or fields.

Project

• Standard library only: zero third-party modules.
• cmd/interpres-decode, a toml-test harness adapter (TOML on stdin, tagged JSON on stdout).
• A runnable example, a table-driven Go test suite, and a canonical just recipe set whose gates recipe is the definition of done.
• Hand-written CI pipelines for the test, race and release gates.
• SECURITY.md for private vulnerability reports.