Skip to content

PaperRoute Tracker v1.0.1

Latest

Choose a tag to compare

@github-actions github-actions released this 15 Nov 14:56
6af1399

PaperRoute Tracker v1.0.1 — Certification Follow-ups

Local-first, privacy-focused app for tracking academic manuscripts from draft through submission, revision, and publication.

This is PaperRoute 1.0.1, a patch for v1.0.0. It fixes six ways to lose people, affiliations or typed text on the manuscript page, or to have unsaved edits written early, each present since v0.9.2 or earlier. All six were found on November 14, 2026, the day v1.0.0 was released, by reading the code; none is new in 1.0. It also takes what 1.0's certification noted. It adds no window, and your library stays on Schema 9. It follows the v1.0.1-rc.1 Preview and has its code.

Fixed

  • ORCID works imported from a manuscript page are kept. On a manuscript page's Authors tab, Manage Library opens Authors & Affiliations. An ORCID works import started there, with ORCID..., did two things it should not have. It wrote the page's unsaved author changes to disk without Save, so Discard did not undo them on disk. And the next save of the library dropped the imported works, although "ORCID Import Complete" had said they were imported; they never appeared on the Board. This was present since v0.2.0. An import from Library > Authors & Affiliations was not affected, and neither was an import that applied only an iD, names or affiliations. An import started from the manuscript page now leaves the page's unsaved changes unsaved, and its imported works are kept and shown on the Board (PR #355).
  • If you lost works this way, import them again from Library > Authors & Affiliations. If you chose Discard after such an import, open that manuscript and check its Authors tab.
  • People a Crossref lookup added are kept until Save. On a manuscript page, DOI & Crossref Metadata... can add new people and affiliations, which wait until you save the page. If Submission Readiness..., Manage Library or Journal, Preprint & Links... (closed with Save & Close) was opened before Save, they were lost. The manuscript then listed them as "[Missing author record]", and no message appeared. This was present since v0.2.0, and through Submission Readiness since v0.4.0. People the lookup matched to someone already in your library were not affected, and neither was a lookup saved before any of those windows was opened. The people and affiliations a lookup added are now kept when any of those windows is opened before Save: Save writes them, and Discard drops them (PR #355).
  • If a manuscript shows "[Missing author record]" this way, remove those entries on its Authors tab and run the Crossref lookup again. The names and iDs the lookup supplied were never saved, so no backup holds them.
  • A BibTeX or RIS import keeps the people you just saved. On a manuscript page, Manage Library opens Authors & Affiliations, which saves each change at once. If you added or changed people, ORCID iDs or affiliations there and then left the page without its own Save (with nothing to save, after Discard, or by answering No when asked to save changes), Import BibTeX / RIS... saved the main window's older copy of the author library over them. They were lost, and the import created those people again under new ids. This was present since v0.6.0. Changes made from Library > Authors & Affiliations were not affected, and neither were changes followed by the page's own Save. The import now reads the author library from disk first, and imports nothing if it cannot read it (PR #363).
  • A person the saved manuscript names can no longer be deleted from its page. On a manuscript page, after you removed an author on the Authors tab without saving, Manage Library let you delete that person, or one of their affiliations, at once. Discard then left the saved manuscript naming a deleted record: for a person, its Authors tab showed "[Missing author record]" and the name, ORCID iD and affiliations were gone; for an affiliation, it was gone from the author. This was present since v0.2.0. A delete from Library > Authors & Affiliations was not affected. From the page, such a delete is now refused while the saved manuscript still names the person or affiliation, and the message says the manuscript open on the page counts as saved until you save its changes (PR #363).
  • If people, iDs or affiliations went missing in either way, authors.bak, beside authors.json in the data folder, is the author library as it was before its last save, so it still holds them if nothing has saved the author library since. A BibTeX or RIS import also put a copy of both files in the backups folder before it ran. The User Guide's Recovering your library says where these are. Otherwise restore a backup, or add the people again.
  • A recovered draft keeps the people a Crossref lookup added. When a session ends without asking, for example when Windows shuts down, PaperRoute keeps a recovery draft of the open manuscript page and offers it under "Recovered Changes" at the next start. The draft held the manuscript, but not the people and affiliations DOI & Crossref Metadata... had added and that were waiting for Save. So a draft recovered after such a lookup named people your author library did not have: its Authors tab showed "[Missing author record]", and the names, ORCID iDs and affiliations the lookup had supplied were gone. This was present since v0.9.2, when the recovery draft was added. The draft now also holds the people, affiliations and tag colours the page had not saved, and a recovered draft puts them back as unsaved changes: Save writes them, and Discard, or answering No to "Recovered Changes", drops them (PR #366).
  • If a recovered draft showed "[Missing author record]" this way, remove those entries on the manuscript's Authors tab and run the Crossref lookup again. A draft written by v1.0.0 or earlier never held those people, so 1.0.1 cannot bring them back from it, and no backup holds them.
  • What you type on the manuscript page is kept through windows and steps that leave it alone. On a manuscript page's Overview, a Title or Target journal you typed, or a Current stage you chose, and had not saved, could be silently replaced by the saved value, and Save then wrote the old value. It happened after DOI & Crossref Metadata... with Title left unticked; after a submission or decision step (Add, Edit or Delete a submission, a revision deadline, or Save's own Editorial Decision Required prompt); and after Journal, Preprint & Links... closed with Save & Close. No message appeared. This was present since v0.2.0, and after a submission or decision step since v0.3.0. Now a window or step puts its value in one of those boxes only when it changes that field itself, or when you choose the value there; otherwise what you typed or chose stays, and Save writes it (PR #367).
  • What still replaces typed text. A window or step that itself changes the field puts its value in the box, over what you typed there: for example a submission to another journal, a decision that moves the stage, or deleting a submission so that the target journal falls back to an earlier one. So does a value you choose in a window: a journal made the target from a shortlist; another journal's name, or a journal picked from the library, in Journal, Preprint & Links..., even when its name is the one saved; or DOI & Crossref Metadata... with Title ticked, even when Crossref's title is the one saved. A step that sets the value the manuscript already had leaves your text alone: a new submission or decision at the journal already saved as the target keeps a target journal you typed, and Save writes the typed one. Two cases still behave as in v1.0.0, and are planned for v1.1 (#368). A target journal you typed before or after a decision recorded or deleted in a submission's details on the Submissions tab is replaced at the next step that refreshes the boxes, such as adding or editing a submission or Journal, Preprint & Links... closed with Save & Close, because those details change the target journal without updating the Target journal box. And a stage you chose right after a Save that recorded a submission or decision for another stage is replaced at the next step.
  • If a box showed the saved value again in an earlier version and you chose Save, the old value was written: type your text again and save.

Changed

  • The Board's stage and sort filters. Each step through the open stage list redrew the Board, which held the window on a large library. The Board now redraws for a stage or sort choice once, when the open list closes, with the shelves' drawing held while it does. A step with the arrow keys on the closed box still redraws at once. On a 500-manuscript library, in the test suite, three steps through the open list and then closing it took 392 to 583 ms, against 988 to 2,227 ms before, and one change from All stages to Submitted took 199 to 401 ms, against 555 to 768 ms (PR #356).
  • The first online request after a resume. After a virtual machine resumes from a saved state, the first online request could fail with "couldn't reach", as an update check did in 1.0's certification: PaperRoute's pool still held connections the network had dropped. A computer whose network drops connections without notice may meet the same. When a request that only reads and carries no content, such as an update check, gets no answer at all, PaperRoute now sends it once more, through the same Online services checks. A request with content, such as an AI assistant's text, is never sent twice. Whether one more try is enough after a real resume is under Known limits.
  • If your studies were on OSF. A new User Guide section, If your studies were on OSF, is for researchers who kept a study's plan, materials or data in an OSF project. It says what changes on OSF and when, with each fact linked to the Center for Open Science's own pages, what PaperRoute can hold for a study today, and the steps to take first.
  • Behind the scenes: the repository holds WinGet manifest files for v1.0.0, under packaging/winget, checked by two tests; nothing has been submitted, and PaperRoute is not in WinGet's catalog. The official RO-Crate validator, roc-validator 0.12.1, found no REQUIRED failures in the RO-Crate metadata of a packet export PaperRoute wrote; the record has the commands and counts. Two keyboard tests that failed while someone was using the PC now release the mouse buttons and modifier keys in the test thread's key state before they post a key, and a new test checks that each list that scrolls sideways measures again when its rows change.

Known limits

  • A recovery draft opened by an older version. If v0.9.2 to v1.0.0 opens a recovery draft written by 1.0.1, it recovers the manuscript without the people the draft holds, as those versions always did; a test shows this with a copy of v1.0.0's reader.
  • The first online request after a resume. 1.0.1 does not claim that one more try is enough after a real resume (#339).

Code signing

PaperRoute 1.0.1 is not code-signed. The project's application to the SignPath Foundation for free code signing was declined on October 31, 2026, for want of signs of public adoption, and a new application is welcome later (#216). That issue, planned for v1.1, holds the routes to try next: WinGet, a Microsoft Store (MSIX) build, or a new application once the project has external references. Until releases are signed, your browser and Windows may warn before Setup runs; the User Guide's Installing section explains the warning and how to check the download against SHA256SUMS.txt. The code signing policy says what will be signed, how, and who approves each signing request.

Before upgrading

1.0.1 uses the same Schema 9 library as v0.9.0 to v1.0.0 and migrates no data. It changes no library file's layout, so v1.0.0 can still open a library 1.0.1 has used. The recovery draft, which PaperRoute keeps in the data folder's recovery folder only when a session ends without asking, gains one optional part, which older versions skip (see Known limits). Updating from v1.0.0 or v0.9.2 asks about unsaved edits first. From v0.9.0 or v0.9.1, save any open manuscript page before you choose Download & Restart: those versions do not ask, so the edits would be lost. If you tried the 1.0.1 Preview, an install on the Stable channel is offered v1.0.1; while the channel is Preview, the Stable release is not offered, so choose Stable again under Settings → Preferences... → Update channel. An install made with the Preview's Setup starts on the Stable channel. If PaperRoute runs on more than one computer that shares a Documents folder, update every one of them before saving on any of them. The v1.0.0 release notes describe what you receive with 1.0.1 if you are updating from v0.9.2. The upgrade notes have the details.

Verification

Every automated test passes, with none skipped. Certification is recorded in #352, scoped by what changed since v1.0.0. v1.0.1 has the code of v1.0.1-rc.1, its Preview candidate; only its version and its documents differ. Before the Preview's publication: the release candidate, and CI on its release commit. On the same code, before v1.0.1's release: the speed budgets on a quiet computer, and the Board's stage and sort filters by keyboard, by mouse and with Narrator. Then the published-asset audit, and the installed-app checks on the Preview's published build, on a Windows 11 test computer: what Edge and Windows show for Setup; five clean installs; the live update from v1.0.0 with byte-identical data files; the first update check after the test computer resumes; each of the six fixes followed by hand, with the library files compared before and after, and the recovery draft's fix by restarting Windows with a Crossref lookup unsaved; a recovery draft exchanged with v1.0.0 both ways; the core route walked by hand; and the new User Guide section in Help. Checks that 1.0.1 does not touch carry over from 1.0's certification (#313), each recorded in #352 with its reason. For v1.0.1 itself: the release candidate and CI on its release commit, then the published-asset audit and, on its published build, what Edge and Windows show for Setup, five clean installs, the live updates from v1.0.0 and from v0.9.2 on the Stable channel, and what an install of the 1.0.1 Preview is offered.