`. Setext headings are deliberately not' . ' supported — `---` collides with DokuWiki HR and `===` would' . ' collide with DokuWiki heading syntax.', 30 => 'thematic break vs. list-item precedence (`* * *` between list' . ' items): requires internal-space HR support and a GfmListblock' . ' guard so the list refuses to absorb the HR-shaped line. Both' . ' out of scope; the line stays a list-item body.', 31 => 'thematic break inside list with different bullet (`- * * *`):' . ' depends on internal-space HR support inside the sub-parsed' . ' item body. See example 30.', // -------------------------------------------------------------------- // Fenced code blocks (GfmCode / GfmFile) — deliberate simplifications // versus strict GFM. All of these are consequences of lexer constraints // (no regex backreferences) or the deliberate column-0-only policy. // -------------------------------------------------------------------- 94 => 'fenced code: closing fence must be ≥ opening length — DokuWiki' . ' accepts any 3+ run as a closer (no regex backreferences for' . ' length pairing). Deliberate relaxation.', 95 => 'fenced code (tilde variant): closing fence must be ≥ opening' . ' length — see example 94.', 96 => 'fenced code: unclosed fence — DokuWiki convention requires a' . ' closer (matches DW tag), so unclosed fences stay' . ' literal rather than consuming to EOF. GFM spec rule depends' . ' on CommonMark\'s two-pass block parser, which our single-pass' . ' lexer cannot implement fully anyway (see example 98).', 97 => 'fenced code: unclosed fence with intervening short run — stays' . ' literal, see example 96.', 101 => 'fenced code: opener indented 1 space — DokuWiki requires' . ' column-0 fences. Indent tolerance + per-line body dedent out' . ' of scope.', 102 => 'fenced code: opener indented 2 spaces — see example 101.', 103 => 'fenced code: opener indented 3 spaces — see example 101.', 105 => 'fenced code: closer indented 2 spaces — column-0-only policy,' . ' see example 101.', 106 => 'fenced code: indented opener with less-indented closer —' . ' column-0-only policy, see example 101.', 107 => 'fenced code: 4-space-indented closer — with column-0-only' . ' policy there is no valid closer, so the fence stays literal' . ' (see example 96).', 109 => 'fenced code: malformed closer `~~~ ~~` (space-broken run) —' . ' with no valid closer the fence stays literal (see example 96).', 108 => 'fenced code: `` `` is not a valid fence; GFM falls back to an' . ' inline code span of length 3. Inline spans with n≥3 not' . ' implemented (GfmBacktickSingle/Double cover only n=1, n=2).', 111 => 'fenced code interrupting Setext heading (`foo\n---`): Setext' . ' headings are deliberately not supported — the `---` underline' . ' collides with DokuWiki\'s horizontal rule and `===` would' . ' collide with DokuWiki heading syntax.', 115 => 'fenced code: `` `` backtick-fence-with-backticks-in-info-string' . ' is invalid; GFM falls back to n=3 inline span — inline spans' . ' with n≥3 not implemented. See example 108.', // -------------------------------------------------------------------- // Code-span edge cases that collide with project-wide decisions // (no raw HTML, no GFM angle-bracket autolinks, typography on by // default) or with the single-pass lexer's limits. // -------------------------------------------------------------------- 351 => 'code span vs. emphasis: cross-positional precedence would require' . ' a pre-scan pass — the single-pass lexer matches leftmost-first' . ' and cannot reject an earlier emphasis opener because a later' . ' backtick span would consume its closer', 353 => 'code span: the trailing `"` outside the span is converted to a' . ' curly quote by DokuWiki typography, diverging from the spec HTML', 354 => 'raw HTML tag pass-through: DokuWiki does not render raw HTML by' . ' default; `` blocks are the opt-in', 356 => 'GFM angle-bracket autolink ``: not implemented — we' . ' rely on DokuWiki\'s existing bare-URL detection, which does not' . ' parse `` form', // -------------------------------------------------------------------- // CommonMark §6.2 flanking-delimiter analysis — deliberately not // implemented. DokuWiki's regex lexer uses leftmost-match and cannot // apply CommonMark's left/right-flanking rules that distinguish // word-chars, whitespace, and punctuation for `*`/`_` delimiters, or // the "multiple-of-3" rule for overlapping runs. These examples all // rely on that machinery. // -------------------------------------------------------------------- // Unicode whitespace in flanking context. Our `\s` is ASCII-only // because the lexer doesn't set the PCRE `u` flag. 363 => 'Unicode whitespace (U+00A0) flanking — requires u-flag-aware regex', // Punctuation-adjacent flanking for `*` / `_` / `**` / `__` 362 => 'flanking: punctuation-adjacent `*` (left-flanking vs. right-flanking)', 368 => 'flanking: punctuation-adjacent `_`', 372 => 'flanking: intraword `_` with punctuation inside', 377 => 'flanking: `*` followed by `(` requires punctuation-aware flanking', 378 => 'flanking: nested `*(*foo*)*` requires flanking + balanced-pair analysis', 382 => 'flanking: nested `_(_foo_)_` requires flanking + balanced-pair analysis', 389 => 'flanking: punctuation-adjacent `**`', 394 => 'flanking: punctuation-adjacent `__`', 401 => 'flanking: `**` followed by `(`', 404 => 'flanking: nested `*bar*` inside `**foo ... foo**` with punctuation', 407 => 'flanking: `__` followed by `(`', 470 => 'flanking: nested `*_foo_*` requires balanced-pair analysis', 472 => 'flanking: nested `_*foo*_` requires balanced-pair analysis', // Intraword `__` strong (even multibyte) — flanking rule for `_` requires // examining whether the delimiter run is word-boundary-flanking, which our // simple lookbehind/lookahead approximation doesn't fully match. 395 => 'flanking: intraword `__` (`foo__bar__`) — left-flanking vs right-flanking', 396 => 'flanking: intraword `__` across digits (`5__6__78`)', 397 => 'flanking: intraword `__` with Cyrillic', 398 => 'flanking: `__foo, __bar__, baz__` — flanking + balanced pairing', 409 => 'flanking: `__foo__bar` — intraword close', 410 => 'flanking: intraword `__` with Cyrillic (leading)', 411 => 'flanking: `__foo__bar__baz__` — multiple `__` pairs with flanking', 412 => 'flanking: `__(bar)__.` — punctuation-adjacent', // Overlapping / multiple-of-3 rule for runs 416 => 'CommonMark rule 9 (overlapping same-delimiter `_foo _bar_ baz_`)', 417 => 'CommonMark overlapping `_` / `__` with flanking', 418 => 'CommonMark overlapping `*foo *bar**` — multiple-of-3 rule', 419 => 'CommonMark nested `*foo **bar** baz*` — balanced-pair analysis', 421 => 'CommonMark overlapping `*foo**bar*` — multiple-of-3', 422 => 'CommonMark nested `***foo** bar*` — triple-delimiter analysis', 423 => 'CommonMark nested `*foo **bar***` — triple-delimiter analysis', 424 => 'CommonMark nested `*foo**bar***` — triple-delimiter analysis', 425 => 'CommonMark triple `foo***bar***baz` — triple-delimiter analysis', 426 => 'CommonMark long delimiter runs `foo******bar*********baz`', 427 => 'CommonMark deeply nested `*foo **bar *baz* bim** bop*`', 434 => 'CommonMark overlapping `__foo __bar__ baz__` — multiple-of-3', 435 => 'CommonMark `____foo__ bar__` — leading long delimiter run', 436 => 'CommonMark `**foo **bar****` — trailing long delimiter run', 439 => 'CommonMark nested `***foo* bar**` — triple-delimiter', 440 => 'CommonMark nested `**foo *bar***` — triple-delimiter', 441 => 'CommonMark deeply nested `**foo *bar **baz** bim* bop**`', // `__foo_` / `_foo__` — mixing `_` and `__` requires flanking to decide // which delimiter pairs open/close. 463 => 'flanking: `__foo_` — mixed `_`/`__` pairing', 464 => 'flanking: `_foo__` — mixed `_`/`__` pairing', 465 => 'flanking: `___foo__` — delimiter-run length analysis', 466 => 'flanking: `____foo_` — delimiter-run length analysis', 467 => 'flanking: `__foo___` — delimiter-run length analysis', 468 => 'flanking: `_foo____` — delimiter-run length analysis', // Long delimiter runs require excess-drop logic (2 outer chars dropped // from each side). Stack-based pairing needed — out of scope. 473 => 'CommonMark `****foo****` — excess-drop (4+4 → strong only)', 474 => 'CommonMark `____foo____` — excess-drop (4+4 → strong only)', 475 => 'CommonMark `******foo******` — excess-drop (6+6 → strong only)', 477 => 'CommonMark `_____foo_____` — excess-drop (5+5 → em+strong, 2 dropped each side)', // Overlapping / crossing delimiters 478 => 'CommonMark `*foo _bar* baz_` — overlapping different delimiters', 479 => 'CommonMark `*foo __bar *baz bim__ bam*` — crossing delimiters', 480 => 'CommonMark `**foo **bar baz**` — overlapping same delimiter', // -------------------------------------------------------------------- // Inline link `[text](url)` — features GfmLink deliberately does not // implement. Either rarely-used syntax paid for with disproportionate // regex complexity, or single-pass-lexer limits that can't be worked // around inside one mode. // -------------------------------------------------------------------- // GFM link title attribute (`"title"` / `'title'` / `(title)` after // the URL). Parses cleanly but is discarded: DokuWiki's link handler // instructions have no title-attribute slot, and plumbing one through // every renderer is out of scope for GfmLink. 493 => 'link title attribute: GfmLink parses but discards — DokuWiki link instructions have no title slot', 513 => 'link title attribute (three quoting styles): discarded by GfmLink', 515 => 'link title separated by non-breaking space: title slot not supported', 516 => 'link title with nested balanced quotes: Markdown.pl quirk, not supported', 517 => 'link title with different quote type for inner quotes: title slot not supported', 518 => 'multi-line link title: title slot not supported', // Pointy-bracket link destinations `<...>`. Rarely used; regex cost // and interaction with raw-HTML detection outweigh the benefit. 496 => 'pointy-bracket link destination `<>`: not supported', 498 => 'pointy-bracket destination with spaces `<...>`: not supported', 500 => 'pointy-bracket destination with newline: not supported', 501 => 'pointy-bracket destination containing `)`: not supported', 502 => 'pointy-bracket destination with trailing backslash: not supported', 503 => 'malformed pointy-bracket destinations: renderer output differs', 507 => 'pointy-bracket destination wrapping unbalanced parens: not supported', // Balanced-parens inside URL destinations. 505 => 'balanced-parens in URL destination: not supported (regex single-level)', // Other URL-level edges. 495 => 'empty URL destination `[link]()`: pattern requires non-empty URL', 510 => 'backslash in URL destination: URL-encoding differs from spec', 511 => 'HTML entity / percent-encoding in URL: renderer normalization differs', 512 => 'link destination that parses as a title: edge case not supported', // Inherent single-pass-lexer limits for link text containing nested // structures. These cannot be resolved inside one mode. 522 => 'nested bracket forms inner link, outer falls back to literal', 526 => 'nested links: inner is a link, outer falls back to literal', 527 => 'nested links inside emphasis: not supported', 529 => 'link text grouping vs. emphasis: leftmost-match cannot override', 530 => 'emphasis/bracket crossing: leftmost-match cannot override', 532 => 'raw HTML inside link text: project-wide "no raw HTML" limit', 533 => 'code span inside link text: requires pre-scan pass (see #351)', 534 => 'autolink inside link text: raw `` autolinks not supported (see #356)', // Reference links (`[text][id]`, `[text][]`, `[foo]` with matching // `[foo]: url` definition). Not implemented: resolving forward // references would require a two-pass parse, but DokuWiki's lexer is // single-pass. Inline links `[text](url)` are the only supported // form. 535 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 536 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 537 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 538 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 539 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 540 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 541 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 542 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 543 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 544 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 545 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 546 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 547 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 548 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 549 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 550 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 551 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 552 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 553 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 557 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 558 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 560 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 561 => 'collapsed reference link: forward-reference definitions not supported (single-pass lexer)', 562 => 'collapsed reference link: forward-reference definitions not supported (single-pass lexer)', 563 => 'collapsed reference link: forward-reference definitions not supported (single-pass lexer)', 564 => 'collapsed reference link: forward-reference definitions not supported (single-pass lexer)', 565 => 'shortcut reference link: forward-reference definitions not supported (single-pass lexer)', 566 => 'shortcut reference link: forward-reference definitions not supported (single-pass lexer)', 567 => 'shortcut reference link: forward-reference definitions not supported (single-pass lexer)', 568 => 'shortcut reference link: forward-reference definitions not supported (single-pass lexer)', 569 => 'shortcut reference link: forward-reference definitions not supported (single-pass lexer)', 570 => 'shortcut reference link: forward-reference definitions not supported (single-pass lexer)', 571 => 'shortcut reference link with escape: forward-reference definitions not supported (single-pass lexer)', 572 => 'shortcut reference link with emphasis: forward-reference definitions not supported (single-pass lexer)', 573 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 574 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 575 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 576 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 577 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 578 => 'reference link: forward-reference definitions not supported (single-pass lexer)', 579 => 'reference link: forward-reference definitions not supported (single-pass lexer)', // -------------------------------------------------------------------- // Inline image ``. The XHTML renderer's default media // rendering diverges from GFM's bare (it wraps in a details // with fetch.php/detail.php proxy URLs) — GfmSpecTest uses // SpecCompatRenderer to emit spec-shape bare , so only the // parser-level or feature-level gaps remain as skips: title attribute // (no DW slot), reference images, pointy-bracket destinations, nested // brackets, and escape-dependent cases. // -------------------------------------------------------------------- 580 => 'image with title attribute: GfmMedia discards titles (no DW slot)', 581 => 'reference-style image: forward-reference definitions not supported (single-pass lexer)', 582 => 'nested image-in-image `](y)`: alt class forbids brackets;' . ' leftmost-match cannot reorder — outer falls back to literal (see #526)', 583 => 'link-in-image alt `](y)`: alt class forbids brackets;' . ' leftmost-match cannot reorder — outer falls back to literal (see #526)', 584 => 'collapsed reference-style image: forward-reference definitions not supported', 585 => 'full reference-style image: forward-reference definitions not supported', 587 => 'image with title attribute: title discarded (no DW slot)', 588 => 'pointy-bracket image destination `![alt]()`: not supported (see GfmLink #496)', 590 => 'reference-style image: forward-reference definitions not supported', 591 => 'reference-style image (case-insensitive label): forward-reference definitions not supported', 592 => 'collapsed reference-style image `![foo][]`: forward-reference definitions not supported', 593 => 'collapsed reference-style image with emphasis in label: forward-reference definitions not supported', 594 => 'collapsed reference-style image (case-insensitive): forward-reference definitions not supported', 595 => 'reference-style image with intervening whitespace: forward-reference definitions not supported', 596 => 'shortcut reference-style image `![foo]`: forward-reference definitions not supported', 597 => 'shortcut reference-style image with emphasis: forward-reference definitions not supported', 598 => 'image with unescaped nested brackets `![[foo]]`: literal-fallback behavior not supported', 599 => 'shortcut reference-style image (case-insensitive): forward-reference definitions not supported', // -------------------------------------------------------------------- // ATX heading collisions with DokuWiki-specific behavior. // -------------------------------------------------------------------- 38 => 'ATX heading with leading spaces: GFM tolerates 0-3 spaces of' . ' indent before the opener; we require the `#` at column 0.' . ' Indent tolerance collides with DokuWiki\'s 2-space-indent' . ' preformatted block and isn\'t worth untangling', 39 => 'indented code block: DokuWiki uses 2-space indent for' . ' preformatted; GFM 4-space indented code blocks are not' . ' implemented', 40 => 'indented code block: 4-space indent after a paragraph is a' . ' continuation in GFM but preformatted in DokuWiki — not' . ' implemented', 41 => 'ATX heading with leading spaces: second heading is indented' . ' by 2 spaces; we require the `#` at column 0', 49 => 'empty ATX heading: DokuWiki\'s XHTML renderer deliberately' . ' skips blank headings (blank() guard in Doku_Renderer_xhtml::header)', // -------------------------------------------------------------------- // List items / Lists — list features GfmListblock deliberately does // not implement. The simplifications are by design: indentation uses // a fixed 2-space-multiple step starting at 0, lazy continuation is // not supported, and the rewriter groups items by 'u'/'o' type only. // The buckets are: // // A. Extra spaces after the marker. CommonMark rolls them (up to // 4) into the content column; we dedent at `marker_width + 1`, // collapsing the extras. // B. 1- or 3-space indent for nesting (we round down to nearest 2). // C. Lazy continuation (column-0 paragraph wrap inside an item). // D. Strict CommonMark loose/tight classification (every blank line // between items / inside items reclassifies; we use a simpler // single-paragraph-tight, multi-paragraph-loose rule). // E. Marker-character-change splits ordered lists ('.' vs ')') or // unordered ('-' vs '+' vs '*'). Our rewriter groups by 'u' / 'o' // type only, not by marker character. // F. List interrupting a paragraph without a blank line — requires a // multi-pass block parser to revisit prior text. // // Examples that depend on a pending mode (GfmQuote, GfmEscape, …) are // intentionally NOT skipped — they remain visible failing tests until // the mode lands. // -------------------------------------------------------------------- // -------------------------------------------------------------------- // Block quotes — deliberate scope reductions vs. strict GFM. The // unified GfmQuote mode (replacing DW Quote) covers `>` blockquotes // for both DW and MD pages, but several CommonMark blockquote rules // are out of scope: // // - 1-3 space indent before `>` (column-0-only policy, consistent // with GfmCode / GfmFile / GfmHeader). // - Lazy continuation (paragraph text without `>` on continuation // lines). Same policy as GfmListblock — markers required on // every line. // - Headers inside quotes — sub-parser excludes BASEONLY so header // instructions don't drive TOC/section-edit anchors that don't // compose with `

`. Same rationale as GfmListblock's // header exclusion inside list items. // - Setext-style block constructs (the `---` underline collides // with DW's HR rule). // // Examples that depend on still-pending modes (GfmHr) are // intentionally NOT skipped — they stay visible until those modes // land. // -------------------------------------------------------------------- 206 => 'block quotes: header inside quote — sub-parser excludes' . ' BASEONLY (TOC / section-edit anchors do not compose with' . ' `

`). Same policy as GfmListblock for `

`.', 207 => 'block quotes: header inside quote with no space after `>` —' . ' see #206 for the BASEONLY exclusion rationale.', 208 => 'block quotes: leading-space `>` (1-3 spaces of indent) —' . ' column-0-only policy, consistent with GfmCode / GfmFile.', 210 => 'block quotes: lazy continuation `> # Foo\n> bar\nbaz` —' . ' every quote line must begin with `>` at column 0. Same' . ' policy as GfmListblock.', 211 => 'block quotes: lazy continuation `> bar\nbaz\n> foo` —' . ' see #210.', 212 => 'block quotes: Setext heading underline `---` after `> foo`' . ' — no Setext headings (the `---` collides with DW HR syntax).', 215 => 'block quotes: fenced code block split across blockquote' . ' boundary — fence inside quote followed by non-`>` lines' . ' depends on the same lazy-continuation rule we do not' . ' implement (see #210).', 216 => 'block quotes: lazy continuation `> foo\n - bar` — see #210.', 225 => 'block quotes: lazy continuation `> bar\nbaz` — see #210.', 227 => 'block quotes: lazy continuation `> bar\n>\nbaz` — see #210.', 228 => 'block quotes: lazy continuation in nested quote' . ' `> > > foo\nbar` — see #210.', 229 => 'block quotes: lazy continuation across nested levels' . ' `>>> foo\n> bar\n>>baz` — see #210.', 232 => 'list items: marker-width content-column alignment (A)', 235 => 'list items: marker-width content-column alignment (A)', 249 => 'list items: marker-width-driven content-column alignment for `10. foo` (A)', 254 => 'list items: marker-width content-column alignment edge case (A)', 258 => 'list items: marker-width content-column for `1. foo` (A)', 263 => 'list items: indent ambiguity at column 0/1/2 (B)', 264 => 'list items: 1-space-indent variation (B)', 265 => 'list items: marker-width with multi-line continuation (A)', 266 => 'list items: marker-width with multi-line continuation (A)', 267 => 'list items: lazy continuation (C)', 268 => 'list items: lazy continuation (C)', 270 => 'list items: lazy continuation across blank line (C+D)', 273 => 'list items: list interrupting a paragraph without blank line (F)', 275 => 'list items: 3-space indent rounds to 2 — sub-list under previous item (B)', 276 => 'list items: marker-width content-column with mixed types (A+E)', 277 => 'list items: nested markers on a single line (A)', 278 => 'list items: marker-character switch splits the list (E)', 281 => 'lists: marker-character change splits unordered list `-` -> `+` (E)', 282 => 'lists: ordered delimiter switch splits list `.` -> `)` (E)', 284 => 'lists: list interrupting paragraph without blank line (F)', 286 => 'lists: marker-width content-column alignment for ordered list (A)', 287 => 'lists: triple blank line + indented continuation in deeply nested item (D)', 288 => 'lists: marker-character change at deeper level (E)', 289 => 'lists: marker-character change with type switch (E)', 290 => 'lists: 1-space-indent variations of items, all stay top-level (B)', 291 => 'lists: 1-space-indent variations on ordered list (B)', 292 => 'lists: marker-character change splits inside nested list (E)', 293 => 'lists: marker-character change with mixed indent (E+B)', 294 => 'lists: lazy continuation across types (C+E)', 295 => 'lists: lazy continuation in nested list (C)', 296 => 'lists: lazy continuation across blank line (C+D)', 297 => 'lists: blank-line classification for loose/tight in nested list (D)', 298 => 'lists: blank-line classification (D)', 300 => 'lists: blank-line classification with marker change (D+E)', 301 => 'lists: blank-line classification + marker-width alignment (D+A)', 304 => 'lists: blank line between sub-list items affects loose/tight (D)', 305 => 'lists: blank line between deeply nested items (D)', 306 => 'lists: blank line at the end of a loose list affects classification (D)', // -------------------------------------------------------------------- // Backslash-escape examples (§6.1) that fail for reasons unrelated to // GfmEscape itself: renderer divergences, typography conversion, and // already-skipped GFM features (autolinks, raw HTML, reference links, // discarded link titles). The escape mechanic itself works. // -------------------------------------------------------------------- 308 => 'backslash escapes: apostrophe is rendered as `'` by DW while' . ' the spec expects a literal `\'` — renderer policy difference,' . ' not an escape bug', 310 => 'backslash escapes: DW typography converts straight `"..."` to curly' . ' quotes when $conf[typography] is on, diverging from spec output', 316 => 'backslash escapes inside angle-bracket autolinks: GFM autolink' . ' `` form not implemented (see example 356)', 317 => 'backslash escapes inside raw HTML: raw HTML pass-through is not' . ' supported by default (see example 354)', 318 => 'backslash escapes in link title: title attribute is discarded — DW' . ' link instructions have no title slot', 319 => 'backslash escapes in reference-link definition: link reference' . ' definitions not supported (single-pass lexer cannot resolve' . ' forward references)', ];