Core structuring tags (part/chapter/itemize/item/em/cite-detail) render as raw markup; TOC shows unresolved page references — regression since 2026.3.2, still present in 2026.3.6
Environment
- App: LiiiSTEM / Mogan
- First observed: any release after 2026.3.2
- Still present in: 2026.3.6
- OS: macOS 14.8.9 (Apple Silicon, M3)
~/Library/Caches/MoganLab fully cleared and app fully restarted before testing.
Description
Opening an existing, larger .tm document (a book-length "Companion" manuscript, style chain includes several custom packages alongside base styles), two things go wrong at once:
- The table of contents shows unresolved
?? / ? in place of page numbers for every single entry (Abstract, List of figures, every chapter and section, ...).
- Core structuring and inline tags that belong to the base TeXmacs style — not to any custom package — render as literal unexpanded markup instead of typeset output:
⟨part|Foundations⟩, ⟨prologue⟩, ⟨itemize|, ⟨item⟩, ⟨chapter|Understanding Unicode⟩, ⟨section|What is Unicode?⟩, ⟨em|characters⟩, ⟨cite-detail|Unicode17|§2.1.2, ...⟩.
Screenshots of both symptoms are attached.
How this differs from #4453
#4453 (and its follow-up, see the linked comment in this repository) is scoped to one custom package's nested use-package cascade failing to resolve — the symptom there is limited to that package's own macros rendering raw, while core tags and the rest of the document are unaffected.
Here, tags that are part of the base style itself are failing to expand, and the TOC's unresolved page numbers look like a downstream effect of the document failing to fully typeset, not an independent bug.
What we've tried
We built a minimal, isolated repro to try to reproduce this (see test_02.tmu attached to the companion report on #4453) — that minimal file uses one of the same custom packages (unicode-core) and shows only that package's own macros failing; core tags (chapter, itemize-dot, item) render fine in it. So whatever triggers the broader collapse described here is not reproduced by a single failed nested use-package in isolation — something specific to the full document's larger style/package stack appears to be involved, but we have not yet isolated a small reproduction of it.
Status
Not yet minimized. We're not attaching the full manuscript at this stage since it's an in-progress research document; happy to share its full style/package list, or a redacted excerpt of just the style chain, if that would help triage. Let us know what would be most useful.
Question for the maintainers
Is it plausible that a failed nested use-package resolution somewhere in a longer style/package chain could abort building the environment/DRD for the whole document, rather than degrading gracefully to just that one package's macros? That would explain why the minimal case (single custom package, shallow chain) stays localized, while the full document (deeper chain) loses core tags entirely.
Attachments
- Screenshot: table of contents with unresolved
??
- Screenshot: raw
part/prologue/itemize/item/chapter/section/em/cite-detail markup
Core structuring tags (part/chapter/itemize/item/em/cite-detail) render as raw markup; TOC shows unresolved page references — regression since 2026.3.2, still present in 2026.3.6
Environment
~/Library/Caches/MoganLabfully cleared and app fully restarted before testing.Description
Opening an existing, larger
.tmdocument (a book-length "Companion" manuscript, style chain includes several custom packages alongside base styles), two things go wrong at once:??/?in place of page numbers for every single entry (Abstract, List of figures, every chapter and section, ...).⟨part|Foundations⟩,⟨prologue⟩,⟨itemize|,⟨item⟩,⟨chapter|Understanding Unicode⟩,⟨section|What is Unicode?⟩,⟨em|characters⟩,⟨cite-detail|Unicode17|§2.1.2, ...⟩.Screenshots of both symptoms are attached.
How this differs from #4453
#4453 (and its follow-up, see the linked comment in this repository) is scoped to one custom package's nested
use-packagecascade failing to resolve — the symptom there is limited to that package's own macros rendering raw, while core tags and the rest of the document are unaffected.Here, tags that are part of the base style itself are failing to expand, and the TOC's unresolved page numbers look like a downstream effect of the document failing to fully typeset, not an independent bug.
What we've tried
We built a minimal, isolated repro to try to reproduce this (see
test_02.tmuattached to the companion report on #4453) — that minimal file uses one of the same custom packages (unicode-core) and shows only that package's own macros failing; core tags (chapter,itemize-dot,item) render fine in it. So whatever triggers the broader collapse described here is not reproduced by a single failed nesteduse-packagein isolation — something specific to the full document's larger style/package stack appears to be involved, but we have not yet isolated a small reproduction of it.Status
Not yet minimized. We're not attaching the full manuscript at this stage since it's an in-progress research document; happy to share its full style/package list, or a redacted excerpt of just the style chain, if that would help triage. Let us know what would be most useful.
Question for the maintainers
Is it plausible that a failed nested
use-packageresolution somewhere in a longer style/package chain could abort building the environment/DRD for the whole document, rather than degrading gracefully to just that one package's macros? That would explain why the minimal case (single custom package, shallow chain) stays localized, while the full document (deeper chain) loses core tags entirely.Attachments
??part/prologue/itemize/item/chapter/section/em/cite-detailmarkup