Internally cache network data instead of networks to reduce redundancy in network construction

[maxloeffler]

Prerequisites

• I adhere to the coding conventions (described here) in my code.
• I have updated the copyright headers of the files I have modified.
• I have written appropriate commit messages, i.e., I have recorded the goal, the need, the needed changes, and the location of my code modifications for each commit. This includes also, e.g., referencing to relevant issues.
• I have put signed-off tags in all commits.
• I have updated the changelog file NEWS.md appropriately.
• I have checked whether I need to adjust the showcase file showcase.R with respect to my changes.
• The pull request is opened against the branch dev.

Description

Changelog

This works towards fixing #119.

Make internal network creation functions return network data instead of networks. This should improve performance as it removes the redundancy in the get.author.network, get.artifact.network, and get.commit.network when merging the returns of the internal network creation functions into a single network because merging networks requires decomposition in network data. This decomposition is not necessary anymore. As a consequence, this also removes the necessity to first compose networks before decomposing them directly after.

Added

• Add helper function convert.edge.list.attributes.to.list that converts edge attributes to lists similar to convert.edge.attributes.to.list but that takes an edge list as input instead of a network
• Add helper function construct.network.data that takes vertex-, and edge-data and constructs a network data object while correctly initializing empty input data

Changed / Improved

• Internal network creation functions now return network data instead of networks

[bockthom]

Good job @maxloeffler!

I really appreciate your changes. Hope that it really boosts the performance of multi-network creation. And even if this is not the case, this PR contains numerous changes I am in favor of.

[bockthom]

Looks good to me.

[maxloeffler]

Before I can finish the NEWS.md, I noticed that I forgot the email in the signed-off tag of the first few commit messages. I will add it in so don't wonder for all the commits 😄.

[codecov]

Codecov Report

Attention: Patch coverage is 96.35762% with 11 lines in your changes missing coverage. Please review.

Please upload report for BASE (dev@6d4d726). Learn more about missing BASE report.
Report is 25 commits behind head on dev.

Files with missing lines	Patch %	Lines
util-networks.R	96.35%	11 Missing ⚠️

Additional details and impacted files

@@          Coverage Diff           @@
##             dev     #282   +/-   ##
======================================
  Coverage       ?   82.31%           
======================================
  Files          ?       16           
  Lines          ?     5367           
  Branches       ?        0           
======================================
  Hits           ?     4418           
  Misses         ?      949           
  Partials       ?        0           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[bockthom]

Project coverage is 0.00%.

@maxloeffler Could you please figure out why the upload of the coverage report has failed? After the first failing upload (when Codecov started to report a project coverage of 0.00%) I've re-run the upload and it failed again. Thereafter, I've re-run the Build (latest) in order to generate a new coverage report, but even uploading the new report to codecov failed again – I have no idea why. Codecov only provides us with the following hint:

Unusable report due to issues such as source code unavailability, path mismatch, empty report, or incorrect data format. Please visit our troubleshooting document
for assistance.

As I have no idea what to do here, I'll leave that up to you to investigate this problem... (but it has a low priority; the other tasks you are currently working on are more important).

[bockthom]

Thanks for the fixes regarding directedness. I have few comments on that - merely regarding tests and documentation.

[maxloeffler]

Sorry for the many pushes, im just discovering typos ......

[bockthom]

Sorry for the many pushes, im just discovering typos ......

"such as example multi networks" is also a typo?

[bockthom]

Approved. 🥳
But we need to rerun the CI pipeline as soon as R 4.0 is removed from the pipeline so that we can figure out whether codecov complains about changes in this PR.

[Copilot]

Pull Request Overview

This PR enhances network building performance by caching internal network data rather than repeatedly constructing full network objects. It updates configuration to support new edge attributes and adds directedness enforcement tests.

• Update allowed edge attributes in NetworkConf
• Add tests to enforce directedness rules in author, artifact, and multi-networks
• Refresh changelog to document caching and directedness behavior changes

Reviewed Changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated no comments.

File	Description
util-conf.R	Add 2025 copyright; adjust allowed edge attributes
tests/test-networks.R	Add directedness enforcement tests for partial & multi-networks
NEWS.md	Update changelog entries for caching and directedness changes

Comments suppressed due to low confidence (3)

tests/test-networks.R:1174

• There are directedness tests for author and artifact networks but none for commit networks. Consider adding similar tests for get.commit.network() to ensure commit-network directedness is correctly enforced.

test_that("Enforcement of directedness in sub-networks", {

NEWS.md:14

• [nitpick] The changelog entry references a long list of commit hashes which makes it hard to read. Consider condensing the commit list by referring to the pull request number or grouping hashes, or splitting into multiple bullet points.

- Reduce the amount of redundantly built networks by caching network data internally. This should improve the performance of building multi-networks, especially, when parts of the multi-networks have been built before (#119, PR #282, 64ac42aa743e7f3a724a66bcd551e5b477e30293, ...)

util-conf.R:887

• The update to allowed edge attributes removed 'author.name' and 'author.mail', which are needed when building author-related networks. Please include these attributes back to ensure author metadata isn't dropped.

"event.date", "event.name", "event.info.1", "event.info.2"

[bockthom]

The coverage upload seems to work now again, thanks for the fix @maxloeffler!

Codecov reports 11 lines in util-networks.R that are not covered. Could you please have a look at them?
The first two lines reported can be ignored, codecov complains about two lines of comments not being covered 🙈 The remaining 9 affect either callgraph or commit-interaction networks. Callgraph can also be ignored.
For the commit-interaction networks, @hechtlC and I are not sure whether these lines should be covered by tests or not.
What is your opinion @maxloeffler ? Maybe we need to enhance existing tests that deal with network caching by also testing for cached commit-interaction networks.

[maxloeffler]

@bockthom coronet does not have tests explicitly focused on caching for any network types. Caching is being capitalized on by tests without notice. My opinion is that if you want tests to cover the caching lines for commit interaction networks then we need more regular tests regarding these networks (there are indeed not too many) but this does not concern this PR here.