Reviewer has chosen to be Anonymous

Summary of paper in a few sentences:

The authors provide a very interesting 'historical' perspective piece based on the unforeseen collapse of the OCLC system, and the steps taken to address this sudden gap. They describe their simple and low cost, low maintenance burden solution in some detail. This efficient and simple to use (by adopters) system is based on github and htacess files, and populated through simple YAML files, with appropriate testing/validation throughout the deployment process.

Reasons to accept:

Describes a very popular identifier/resolution system extensively used through the ontological community (OBO). This has significant usage, and would be very useful to developers and actually to end users who wish to learn more. It also provides a historical record on what happened previously when such a system failed, and provides sufficient technical detail to newcomers.

Reasons to reject:

None

Nanopublication comments:

Further comments:

Largely miscellaneous comments/suggestions.

It would be nice to round out this information on what happened to the previously hosted information on purl.org & internet archive. One would imagine that some of the unclaimed purls must have been lost? Were the redirection targets captured for example, even when no longer resolvable? Such information could be useful to may to 'live' targets currently used by the new obo purls.

w3id.org section:
w3id.org.org system typo

Implementation:
3 types of redirect cases are stated. It would be nice to add a concrete example for each if possible..I appreciate there is more information in the 'Testing and deployment' & 'Custom configuration' sections. Could alternatively direct reader to those sections and provide there.

custom configuration
- if possible, it would be interesting to know the total number of 'term browsers' that are used to cover obo ontologies

testing and deployment
- couple of spaces after commas in a few locations
- to automated the -> to automate the
- If all test pass -> If all tests pass

Results
- Is any more information on the 4xx errors available?

Future work
Load balancing and backup/redundancy have been considered. Any thought to moving to entirely cloud-based solutions?
Web-based YAML editor sounds perfect.

By James McCusker https://orcid.org/0000-0003-1085-6059

Review Details

---

Reviewer has chosen not to be Anonymous

Summary of paper in a few sentences:

This paper outlines the methods used to create Persistent URLs (PURLS) for the OBO Foundry project, which manages URI redirection for thousands of ontology classes and over 200 ontologies, after the failure of the free purl.org resource. They present both the method and motivations for their approach, and demonstrate that the approach has resulted in a stable, reliable infrastructure for linked data identifier management.

Reasons to accept:

The paper is readable and the approach, motivation, related work, and evaluation are all described well in the paper. This is a reliable system that is a core part of supporting biomedical ontologies.

Reasons to reject:

Technical Quality:

The authors have not demonstrated reusability by other organizations, either of how an organization might replicate this approach (does this work from the obo files or from OWL ontology files? What is the input, generally, to the yaml files used?), nor does it show an example of it being reused by a second organization. I would prefer to see such an evaluation, and possibly the reuse of the tool by another organization facing similar problems.

Novelty:

While this is a useful contribution in technique, its simplicity makes it difficult to understand its novelty. It seems to be an incremental improvement over w3id, but one that might have been straightforward to discover with similar user cases. It might be helpful to understand what kinds of users have been able to contribute to these configurations - are they generally programmers, data scientists, biologists, or some mixture of that? If it turns out that most semi- and non-technical users are comfortable with this approach, that would be a novel and useful insight even beyond this single application.

Nanopublication comments:

Further comments:

By Alexander Malic https://orcid.org/0000-0002-8394-3439

Review Details

---

Reviewer has chosen not to be Anonymous

Summary of paper in a few sentences:

The decision for selecting Ansible could be clearer. It would be interesting to know how other approaches compare or complement this aproach (e.g. CI tools like Travis-CI, Jenkins or Gitlab Runner; Cron-Jobs that run simple update scripts, full blown webapp with data-base and web-service support) . Does Ansible help make this approach more reusable or was it selected because it fits into an existing infrastructure? What is the risk of using Ansible in regards of maintenance?

Reasons to accept:

It is great to see Dev-Ops methods being used to automate processes. This could inspire others.

Reasons to reject:

Listing to the Ansible implementation script itself is missing.

Nanopublication comments:

Further comments:

Listing 1 shows examples for "exact" and "prefix" but the text below speaks of "exact", "partial" and "regex". It takes some while to understand that it is actually an error and either "prefix" or "partial" is the correct term. Adding a regexp example to Listing 1 would have made it clearer.