Neil Walkinshaw
@neilwalkinshaw.bsky.social
280 followers 93 following 22 posts
Lecturer in Computer Science at the University of Sheffield. Researcher interests in software analysis and testing, and their links to causal reasoning and ML.
Posts Media Videos Starter Packs
neilwalkinshaw.bsky.social
Developed a pesky running injury. Not looking forward to working through the suggestions illustrated by ChatGPT, especially the straight leg raise...
Reposted by Neil Walkinshaw
morphss.bsky.social
New open access book out from our MORPHSS project lead @samuelmoore.org:

"By deploying theoretical literature on science & technology studies, care ethics, & the commons, the book critically interrogates open access & reimagines a more ethical future for researcher-led publishing."

#OA #AHSS
neilwalkinshaw.bsky.social
As with the post office Horizon bug, years of institutional coverup and mismanagement after the bug is exposed. This probably causes more damage than the technical bug itself.
www.bbc.co.uk/news/article...
Courts service accused of covering up IT bug that caused evidence to go missing
Sources say judges in courts and tribunals will have made rulings when evidence was incomplete.
www.bbc.co.uk
neilwalkinshaw.bsky.social
An account of what is possibly the first "monkey tester" (or at least when the term originated) - a program to automatically generate random test inputs. Developed in 1983 to test MacWrite and MacPaint!

www.folklore.org/Monkey_Lives...
Folklore.org: Monkey Lives
www.folklore.org
neilwalkinshaw.bsky.social
I found this very interesting. Could be interesting to consider a variant of this for research paper submissions to Computer Science conferences, where the ballooning review burden is completely unsustainable.

Would need to devise different measures to prevent misuse though.
neilwalkinshaw.bsky.social
AALpy - a very elegant Python library (not by me!) implementing several finite state machine inference algorithms. Includes active and passive approaches, state-merging and observation-table based ones.

github.com/DES-Lab/AALpy
GitHub - DES-Lab/AALpy: An Automata Learning Library Written in Python
An Automata Learning Library Written in Python. Contribute to DES-Lab/AALpy development by creating an account on GitHub.
github.com
neilwalkinshaw.bsky.social
Simple visualisations of code "survival" over time: github.com/erikbern/git...

Several groups of students have tried this on a variety of OSS projects. Anecdotally it seems that the "half-life" of the code for a given version is remarkably short! Would welcome pointers to studies on this!
GitHub - erikbern/git-of-theseus: Analyze how a Git repo grows over time
Analyze how a Git repo grows over time. Contribute to erikbern/git-of-theseus development by creating an account on GitHub.
github.com
neilwalkinshaw.bsky.social
Procedures for testing driverless cars (at least the underlying ML) are still effectively in their infancy, so it's interesting that these will be on UK roads so soon. Particularly interested in accountability (moral & legal) when accidents inevitably occur.

www.bbc.co.uk/news/article...
UK driverless cars unlikely until 2027 - Uber says it's ready now
The previous government had said autonomous vehicles were
www.bbc.co.uk
Reposted by Neil Walkinshaw
hillelwayne.com
Blast from the past: "How Do We Trust Our Science Code?"

Most scientific research code is low quality and riddled with bugs, written by people who aren't professional software developers. And we shouldn't expect them to be!

(Written eight years ago, Jesus)

www.hillelwayne.com/post/how-do-...
How Do We Trust Our Science Code?
In 2010 Carmen Reinhart and Kenneth Rogoff published Growth in a Time of Debt. It’s arguably one of the most influential economics papers of the decade, convincing the IMF to push austerity measures i...
www.hillelwayne.com
Reposted by Neil Walkinshaw
greggay.bsky.social
Reminder: I am hiring 1-2 PhD students to research the use of LLMs to augment human-written test suites!

Interested? Apply here: www.chalmers.se/en/about-cha...
Reach out if you have questions!

#softwaretesting #softwareengineering #softwaredevelopment #LLM #AI #PHD
Vacancies
www.chalmers.se
Reposted by Neil Walkinshaw
greggay.bsky.social
I'm seeking 1-2 Ph.D. students, who will conduct research on how LLMs can extend and augment the software tests created by humans. Please apply, share, and get in touch if you have questions!

Apply at: www.chalmers.se/en/about-cha...
Vacancies
www.chalmers.se
Reposted by Neil Walkinshaw
daviddlevine.com
LLMs hallucinating nonexistent software packages with plausible names leads to a new malware vulnerability: "slopsquatting."
LLMs can't stop making up software dependencies and sabotaging everything
: Hallucinated package names fuel 'slopsquatting'
www.theregister.com
neilwalkinshaw.bsky.social
So impressed by the JOSS publication model. The whole process is run on GitHub! Reviews are submitted as issues! github.com/openjournals...

Who needs big publishers. If there were a way to introduce anonymity into the reviewing process (surely possible), this is the way forward imho.
[REVIEW]: The Causal Testing Framework · Issue #7739 · openjournals/joss-reviews
Submitting author: @jmafoster1 (Michael Foster) Repository: https://github.com/CITCOM-project/CausalTestingFramework Branch with paper.md (empty if default branch): Version: v10.0.1 Editor: @daniel...
github.com
neilwalkinshaw.bsky.social
Our Causal Testing Framework, part of the EPSRC-funded CITCOM project, has just been published in @joss-openjournals.bsky.social.

joss.theoj.org/papers/10.21...

Supports testing of "hard-to-test" systems with large input spaces, nondeterminism and long execution times.
The Causal Testing Framework
Foster et al., (2025). The Causal Testing Framework. Journal of Open Source Software, 10(107), 7739, https://doi.org/10.21105/joss.07739
joss.theoj.org
neilwalkinshaw.bsky.social
Especially short-sighted, given that so many key technologies that underpin the strong growth in the US economy, especially in AI, have emerged from these very research institutions.
neilwalkinshaw.bsky.social
Delighted for my Ph.D. student Richard Somers, whose work on using a digital twin (of blood-glucose dynamics in a human) to test artificial pancreas systems for diabetics has just been published in STVR:

onlinelibrary.wiley.com/doi/epdf/10....
Configuration Testing of an Artificial Pancreas System Using a Digital Twin: An Evaluative Case Study
onlinelibrary.wiley.com