Software Peer Review at rOpenSci is managed by a team of editors. We are piloting a system of a rotating Editor-in-Chief (EiC).
If you’re a guest editor, thanks for helping! Please contact the editor who invited you to handle a submission for any question you might have.
In addition to handling packages (about 4 a year), editors weigh in on group editorial decisions, such as whether a package is in-scope, and determining updates to our policies. We generally do this through Slack, which we expect editors to be able to check regularly.
We also rotate Editor-in-Chief responsibilities (first-pass scope decisions and assigning editors) amongst the board about quarterly.
You do not have to keep track of other submissions, but if you do notice an issue with a package that is being handled by another editor, feel free to raise that issue directly with the other editor, or post the concern to editors-only channel on slack. Examples:
- You know of an overlapping package, that hasn’t been mentioned in the process yet.
- You see a question to which you have an expert answer that hasn’t been given after a few days (e.g. you know of a blog post tackling how to add images to package docs).
- Concerns related to e.g. the speed of the process should be tackled by the editor-in-chief so that’s who you’d turn to for such questions.
- If you’re the EiC or the first editor to respond, assign an editor with a comment of
@ropensci-review-bot assign @username as editor. This will also add tag
1/editor-checksto the issue.
- Submission will automatically generate package check output from ropensci-review-bot which should be examined for any outstanding issues (most exceptions will need to be justified by the author in the particular context of their package.). If you want to re-run checks after any package change post a comment
@ropensci-review-bot check package.
- After automatic checks are posted, use the editor template to guide initial checks and record your response to the submission. You can also streamline your editor checks by using the
pkgreviewrpackage created by associate editor Anna Krystalli. Please strive to finish the checks and start looking for reviewers within 5 working days.
- Check that template has been properly filled out.
- Check against policies for fit and overlap. Initiate discussion via Slack #software-review channel if needed for edge cases that haven’t been caught by previous checks by the EiC. If reject, see this section about how to respond.
- Check that mandatory parts of template are complete. If not, direct authors toward appropriate instructions.
- For packages needing continuous integration on multiple platforms (cf criteria in this section of the CI chapter) make sure the package gets tested on multiple platforms (having the package built on several operating systems via GitHub Actions for instance).
- Wherever possible when asking for changes, direct authors to automatic tools such as
styler, and to online resources (sections of this guide, sections of the R packages book) to make your feedback easier to use. Example of editor’s checks.
- Ideally, the remarks you make should be tackled before reviewers start reviewing.
- If initial checks show major gaps, request changes before assigning reviewers. If the author mentions changes might take time, apply the holding label.
- If the package raises a new issue for rOpenSci policy, start a conversation in Slack or open a discussion on the rOpenSci forum to discuss it with other editors (example of policy discussion).
- Comment with
@ropensci-review-bot seeking reviewers.
- Use the email template if needed for inviting reviewers
- When inviting reviewers, include something like “if I don’t hear from you in a week, I’ll assume you are unable to review,” so as to give a clear deadline when you’ll move on to looking for someone else.
- Assign reviewers with
@ropensci-review-bot assign @username as reviewer.
addcan also be used instead of
to reviewers(plural) instead of
as reviewer(single). The following is thus also valid:
@ropensci-review-bot add @username to reviewers. One command should be issued for each reviewer. If needed later, remove reviewers with
@ropensci-review-bot remove @username from reviewers.
As a (guest) editor, use
- the potential suggestions made by the submitter(s), (although submitters may have a narrow view of the types of expertise needed. We suggest not using more than one of suggested reviewers);
- the Airtable database of reviewers and volunteers (see next subsection);
- and the authors of rOpenSci packages.
When these sources of information are not enough,
- ping other editors in Slack for ideas,
- look for users of the package or of the data source/upstream service the package connects to (via their opening issues in the repository, starring it, citing it in papers, talking about it on Twitter).
- You can also search for authors of related packages on r-pkg.org.
- R-Ladies has a directory specifying skills and interests of people listed.
- You might tweet about the reviewer search.
Your tools for searching reviewers in the Airtable are
- filters (please remove them once you’re done), see example below.
- creating a duplicated view, depending on your Airtable skills.
Have a look at the
last_answer columns before contacting someone!
If someone recently refused because they were busy, it might be best to abstain, whereas someone who refused because of a COI could be contacted again without waiting too long.
For people listed in the Airtable, if you contact them about a review please update the
last_time_contacted column, and enter the category corresponding to their answer in
Only add people to Airtable if they accept to review (otherwise, people should volunteer themselves by filling the Airtable form). Do not enter any information other than GitHub username, name and email address yourself. Point reviewers to the Airtable form.
Here are criteria to keep in mind when choosing a reviewer. You might need to piece this information together by searching CRAN and the potential reviewer’s GitHub page and general online presence (personal website, Twitter).
- Has not reviewed a package for us within the last 6 months.
- Some package development experience.
- Some domain experience in the field of the package or data source
- No conflicts of interest.
- Try to balance your sense of the potential reviewer’s experience against the complexity of the package.
- Diversity - with two reviewers both shouldn’t be cis white males.
- Some evidence that they are interested in openness or R community activities, although blind emailing is fine.
Each submission should be reviewed by two package reviewers. Although it is fine for one of them to have less package development experience and more domain knowledge, the review should not be split in two. Both reviewers need to review the package comprehensively, though from their particular perspective. In general, at least one reviewer should have prior reviewing experience, and of course inviting one new reviewer expands our pool of reviewers.
- Check in with reviewers and authors occasionally. Offer clarification and help as needed.
- In general aim for 3 weeks for review, 2 weeks for subsequent changes, and 1 week for reviewer approval of changes.
- Upon each review being submitted,
- Write a comment thanking the reviewer with your words;
- Record the review via typing a new comment
@ropensci-review-bot submit review <review-url> time <time in hours>. E.g. for the review https://github.com/ropensci/software-review/issues/329#issuecomment-809783937 the comment would be
@ropensci-review-bot submit review https://github.com/ropensci/software-review/issues/329#issuecomment-809783937 time 4.
- If the author stops responding, refer to the policies and/or ping the other editors in the Slack channel for discussion. Importantly, if a reviewer was assigned to a closed issue, contact them when closing the issue to explain the decision, thank them once again for their work, and make a note in our database to assign them to a submission with high chances of smooth software review next time (e.g. a package author who has already submitted packages to us).
- Upon changes being made, change the review status tag to
5/awaiting-reviewer-response, and request that reviewers indicate approval with the reviewer approval template.
@ropensci-review-bot approve <package-name>
- If the original repository owner opposes transfer, add a line with its address to this repos list to ensure the package gets included in rOpenSci package registry.
- Nominate a package to be featured in an rOpenSci blog post or tech note if you think it might be of high interest. Please note in the software review issue one or two things the author could highlight, and tag
- If authors maintain a gitbook that is at least partly about their
package, contact an rOpenSci staff member so they might contact the authors
about transfer to the
The EiC serves for 3 months or a time agreed to by all members of the editorial board. The EiC is entitled to taking scope and overlap decisions as independently as possible (but can still request help/advice). In details, the EiC plays the following roles
Watches all issues posted to the software-review repo (either subscribe to repo notifications on GitHub, or watch the
#software-peer-review-feedchannel on Slack).
Tags issue with
Assigns package submissions to other editors, including self, to handle. Mostly this just rotates among editors, unless the EiC thinks an editor is particularly suited to a package, or an editor declines handling the submission due to being too busy or because of conflicting interests.
@ropensci-review-bot assign @username as editor
Monitors pace of review process and reminds other editors to move packages along as needed.
On assuming EiC rotation, reviews status of current open reviews and reminds editors to respond or update status as needed.
Responds to issues posted to the software-review-meta repo
Makes decisions on scope/overlap for pre-submission inquiries, referrals from JOSS or other publication partners, and submissions if they see an ambiguous case (This last case may also be done by handling editors (see below)). To initiate discussion, this is posted to the rOpenSci Slack editors-only channel along with a small summary of what the (pre-)submitted/referred submission is about, what doubts the EiC has i.e. digesting information a bit. If after one day or two the EiC feels they haven’t received enough answers, they can ping all editors.
- Any editor should feel free to step in on these. See this section about how to respond to out-of-scope (pre-) submissions.
- After explaining the out-of-scope decision, write an issue comment
Requests a new EiC when their rotation is up (set a calendar reminder ahead of your expected end date and ask for volunteers in the editors’ Slack channel)
In some cases online documentation is sparse. Minimal README, no pkgdown website make assessment harder. In that case please ask for more details: even if the package is deemed out-of-scope, the package docs will have gotten better so we are fine asking for these efforts.
Hello <username> and many thanks for your submission. We are discussing whether the package is in scope and need a bit more information. Would you mind adding more details and context to the README? After reading it someone with little domain knowledge should have been informed about the aim, goals and functionality of the package. <optional> If a package has overlapping functionality with other packages, we require it to demonstrate in the documentation [how it is best in class](https://devguide.ropensci.org/policies.html#overlap). Could you add a more detailed comparison to the packages you mention in the README so we can evaluate? </optional>
After discussion with other editors the EiC might invite a guest editor to handle a submission (e.g. if submission volume is large, if all editors have a conflict of interest, if specific expertise is needed, or as a trial prior to inviting a person to join the editorial board).
When inviting a guest editor,
- Ask about conflicts of interest using the same phrasing as for reviewers,
- Give a link to the guide for editors.
If the person said yes (yay!),
- Make sure they enabled 2FA for their GitHub account,
- Invite them to the
ropensci/editorsteam and to the ropensci organization,
- Once they’ve accepted this repo invitation, assign the issue to them,
- Ensure they’re (already) invited to rOpenSci Slack workspace,
- Add their name to the Airtable guest-editor table (so their names might appear in this book and in the software-review README).
After the review process is finished (package approved, issue closed),
- Thank the guest editor again,
- Remove them from the
ropensci/editorsteam (but not from the ropensci organization).
Thank authors for their submission, explain the reasons for the decision, and direct them to other publication venues if relevant, and to the rOpenSci discussion forum. Use wording from Aims and scope in particular regarding the evolution of scope over time, and the overlap and differences between unconf/staff/software-review development.
Reviewers might ask for feedback on e.g. the tone of their review. Beside pointing them at general guidance in this guide, asking editors / opening an issue when such guidance is lacking, here are some review examples that might be useful.
- tough-but-constructive example: the part of this review suggesting a re-write of the vignette: ropensci/software-review#191 (comment).
slopespackage, which ended up being fundamentally redesigned in response to the reviews. All reviews/reviewers were at all times entirely constructive, which seems to have played a major role in motivating the authors to embark on such a major overhaul. Comments such as, “this package does not …” or “has not …” were invariably followed by constructive suggestions for what could be done (there are, for example, several in one of the first reviews).
- tic reviews politely expressed reservations: https://github.com/ropensci/software-review/issues/305#issuecomment-504762517 and https://github.com/ropensci/software-review/issues/305#issuecomment-508271766
- bowerbird useful “pre-review” that resulted in a package split before the actual reviews.
If you are in charge of managing a release of the very book you are reading, use the book release guidance as an issue template to be posted in the dev guide issue tracker, and do not hesitate to ask questions to other editors.
For very small amendments to the dev guide, no PR review is needed. For larger amendments, request review from at least a few editors (if none participated in the discussion related to the amendment, request a review from all of them on GitHub, and in the absence of any reaction merge after a week).
Two weeks before a dev guide release, once the PR from dev to master and the release blog post are ready for review, all editors should be pinged by GitHub (“review request” on the PR from dev to master) and Slack, but the release doesn’t need all of them to explicitly approve the release.
The blog post about a release will be reviewed by editors, and one of
Refer to the general rOpenSci blogging guidance, and the more specific guidance below.
The blog post should mention all important items from the changelog organized in (sub)sections: e.g. a section about big change A, another one about big change B, and one about smaller changes lumped together. Mention the most important changes first.
For each change made by an external contributor, thank them explicitly using the information from the changelog. E.g.
[Matt Fidler](https://github.com/mattfidler/) amended our section on Console messages [ropensci/dev_guide#178](https://github.com/ropensci/dev_guide/pull/178)..
At the end of the post, mention upcoming changes by linking to open issues in the issue tracker, and invite readers to contribute to the dev guide by opening issues and participating in open discussions. Conclusion template:
In this post we summarized the changes incorporated into our book ["rOpenSci Packages: Development, Maintenance, and Peer Review"](https://devguide.ropensci.org/) over the last X months. We are grateful for all contributions that made this release possible. We are already working on updates for our next version, such as ISSUE1, ISSUE2. Check out the [the issue tracker](https://github.com/ropensci/dev_guide/issues/) if you'd like to contribute.