Changelog

Draft any work version from the last one

The headline this week is a new drafting flow for work versions in Curvenote SCMS, cloned from the last non-draft predecessor and cleaned of stale files. These draft improvements are alongside a materially faster PDF figure preview on upload and a new Preview Theme app that renders SCMS articles without first requiring that they be part of a site.

Curvenote SCMS

Draft any work version from the last one. New draft work versions on Curvenote SCMS now clone from the latest non-draft predecessor rather than starting empty or inheriting the whole prior version. Article drafts keep frontmatter metadata but drop inherited files, upload analysis, and preview-thumbnail listings — the selected thumbnail still carries over — so authors start each round on a clean upload surface with a familiar identity. Resume-draft dialogs show v{n} badges on list items so it is obvious which draft belongs to which version.

A new draft work version clones from the latest non-draft predecessor: frontmatter metadata and the selected thumbnail carry over while inherited files are dropped, so authors start each round on a clean upload surface.

Figure 1:A new draft work version clones from the latest non-draft predecessor: frontmatter metadata and the selected thumbnail carry over while inherited files are dropped, so authors start each round on a clean upload surface.

Faster PDF figure previews on upload. Authors uploading a PDF manuscript see a materially faster thumbnail gallery. A new fast path scans up to 99 pages, drops tiny and oversized rasters before rendering, and dedupes repeated paints of the same image, with a PDF-specific busy message and automatic retry when the first pass returns nothing.

A new Preview Theme app for previews. A new Preview Theme renders SCMS submission previews on their own domain. It is a Curvenote-branded landing and a scoped preview pipeline designed for Works that are not yet part of a site. The preview header is a top-nav branded Secure Preview, and the landing links back into SCMS for the corresponding source work — so authors and reviewers can share a working link to a draft without exposing it through the destination site.

Curvenote Sites

Site service accounts announce themselves. When a site admin creates or deletes a site service account, or creates or deletes a token for one, the site emits operational notifications on the shared telemetry channels so account activity is visible alongside the rest of site admin events.

Curvenote Theme

A shared layout scaffolding for preview-style theme apps. The rendering core picks up the primitives from the new Preview Theme — landing and preview-shell components, SCMS-header helpers, and analytics wiring — so future preview surfaces across all Curvenote themes can be composed the same way.

Article listings default to a much larger page. When an articles listing directive omits limit, the listing now defaults to a page of 100 instead of falling back to the SCMS default of 10. Partner sites and readers rendering large catalogues see a full page of results by default rather than the first ten.

Submit a work directly from SCMS

Direct submit-to-site from the work-details page is the headline this week, alongside a new scheduled-jobs and cron admin for SCMS operators and a cleaner upload-and-extract flow for authors. On the reader side, openRxiv Labs picks up sharper citations, references, and abbreviations from a substantive JATS-processing upgrade, and preview pages across Curvenote Theme now render without the site top-nav so drafts look like the article they’ll become.

Curvenote SCMS

Submit a work to a site from work details. The work-details page now includes a version-aware submit control that sends the chosen work version to any eligible site in one step. Sites appear in the picker only when they have an open collection and matching submission kind — or when the work is already on them — so authors don’t see venues they can’t actually submit to. Draft versions stay hidden, and a version that’s already been sent to a site can’t be re-sent by accident.

The version-aware submit control on the work-details page sends a chosen work version to any eligible site in one step, offering only sites with an open collection and matching submission kind.

The version-aware submit control on the work-details page sends a chosen work version to any eligible site in one step, offering only sites with an open collection and matching submission kind.

Cross-version check summaries in the submit bar. When choosing where to submit, the submit bar now shows the most recent run of each check across every non-draft version of the work — so “Not run” only appears when a check has genuinely never been run. When the latest run came from a different version than the one being submitted, a small v{n} badge makes it obvious.

Scheduled jobs. Administrators now get a real cron admin page in SCMS with a name-and-description form, validated HTTP verb and target, and a live schedule preview that shows the next fire time as you edit. Scheduled jobs promote themselves when they’re due, the background queue drain can be paused and resumed for maintenance, and failed check runs now carry retry state that the system can act on.

A cleaner upload metadata flow. Authors uploading works see a per-file document preview with capped scroll, tidy tab titles, and a dedicated state for each file. The authors and affiliations panel gets a simple/edit toggle so the common case reads as a summary and edits are one click away. Metadata auto-extract now only fires on a fresh upload with empty details — not every re-upload — and re-run and clear links are always in reach at both the top and the bottom of the form.

Only offer image checks when there are actually images. The checks form now knows whether an upload contains figures and only offers image-based checks when they apply.

Tighter access control on works. Checks dispatch is now a per-work role granted to owners and contributors instead of an app-wide scope — viewers get read access to check history without being able to run new checks. The whole checks UI is now gated behind a feature flag.

Curvenote Sites

Docs and glue for submit-to-site. The site-side of the new submit-to-site flow ships with an operator-facing walkthrough of how a work moves from SCMS into a site collection, and the site’s submissions loader is updated in step with the SCMS change above.

Curvenote Reader

A shared layout header for preview and article routes. The rendering core now carries an explicit layout-header slot in its layout components, so downstream apps can control the site header from the layout tree instead of re-implementing the same guards inside every header. That’s what makes the Curvenote Theme preview-page change below possible without every reader-family app repeating the work.

openRxiv Labs — Curvenote Reader

In production at openRxiv Labs

openRxiv Labs uses Curvenote Reader for all bioRxiv and medRxiv content — see openRxiv Labs.

Sharper citations, references, and abbreviations. Articles on the openRxiv Reader come through with materially cleaner citations, references, and abbreviations after a substantive upgrade to how JATS content is converted. Superscript citations followed by a separator now parse correctly. Identifiers are pulled out of unstructured citation text so a reference like “Smith 2020” links where it should. Definition lists can serve as abbreviation sources. A MathML bracket-rendering error is fixed and table cross-references (ref-type=tab) now resolve as tables.

Curvenote Theme

Preview pages render without the site top-nav. Preview routes in the standalone Curvenote Theme app no longer show the site top-nav — so authors and reviewers looking at an unpublished draft see the article the way it will read once published, without a global header that will be replaced at publish time.

Reader enhancements database

New create-work flow for My Works that has been released to some customers. The work-details metadata card also gets a cleanup — version DOIs fall back to the work-level DOI, and empty DOI and license fields are hidden instead of showing blank labels. We’re also previewing the enhancements database we’re building for Curvenote Reader — layering equations and abbreviations into view for each version of a paper.

Curvenote SCMS

Create-work flow on My Works. The single “New” button has becomes a create-work menu — Article, Check My Work, and extension-provided options such as PMC deposit — with each choice routed to the right handler instead of defaulting to a plain work creation.

The “Create new submission” menu on My Submissions offers Article, Check My Work, and extension-provided options such as PMC Deposit — each routed to the right handler.

Figure 1:The “Create new submission” menu on My Submissions offers Article, Check My Work, and extension-provided options such as PMC Deposit — each routed to the right handler.

DOIs improvements for Work details. When a version has no DOI of its own, the work-details page now shows the work-level DOI, so the record still displays a resolvable identifier where one exists instead of an empty DOI row. The metadata card hides the DOI row when neither the version nor the work has a DOI, and hides the license row when no license is set — so incomplete records no longer show empty labels.

Curvenote Reader

An enhancements database, layered on demand. We’re building an enhancements database for Curvenote Reader that layers extra context onto content on demand, for each version of a paper — starting with equations (which often arrive as images) and abbreviations. The underlying technology has come together over the last few weeks: a way to attach and render these enhancements on top of the existing content, without changing the source, for every version of a work.

openRxiv Labs — Curvenote Reader

In production at openRxiv Labs

openRxiv Labs uses Curvenote Reader for all bioRxiv and medRxiv content — see openRxiv Labs.

Bringing math into view. A lot of the feedback we hear on Reader is that math comes in as images — a property of the underlying bioRxiv corpus rather than the Curvenote Reader itself. The enhancements work above is aimed squarely at this: layering equations (and tables in future) into view for each version of a preprint, so math and glossaries read cleanly even when the source only provides an image. It is in active development and will roll out to openRxiv Labs as it lands.