Make, Fix, and Test

A few weeks ago, in A Good Tester is All Over the Place, Joep Schuurkes described a model of testing work based on three axes:

• do testing yourself or support testing by others
• be embedded in a team or be part of a separate team
• do your job or improve the system

It resonated with me and the other testers I shared it with at work, and it resurfaced in my mind while I was reflecting on some of the tasks I've picked up recently and what they have involved, at least in the way I've chosen to address them. Here's three examples:

Documentation Generation

We have an internal tool that generates documentation in Confluence by extracting and combining images and text from a handful of sources. Although useful, it ran very slowly or not at all so one of the developers performed major surgery on it.

Up to that point, I had never taken much interest in the tool and I could have safely ignored this piece of work too because it would have been tested by its maintainers and the consumers of its output. However, I decided to review the change precisely because I was unfamiliar with this part of our infrastructure.

When I say this was primarily a learning opportunity for me don't take it the wrong way. I found a nice way to test and I uncovered some interesting issues with the new implementation, but my motivation was to do something new and see what it could teach me. In the absence of other constraints I will often bias towards picking up tasks which give me that.

Schema Validation

We have a pipeline that generates a handful of large JSON files and had been asked to update the format to align with changes to the underlying source data. I had paired with the developer to understand the task before implementation and so had an idea that I would start my testing by dumping the same source content in both old and new formats and diffing them by eye to look for examples of expected changes, then attempting to reverse the new formatting with a simple Python script to check for unwanted changes.

That worked fine, so I turned my attention to validation. The pipeline asserts that the JSON output is valid against a schema and so naturally the schema had changed. I reviewed the edits and they looked like the kinds of changes I might expect ... if I understood the syntax of JSON schemas ... which I do not.

So I decided to go a little further and installed ajv, a tool for using JSON schemas to validate content,  because it supports the version of the schema standard that we had used and because it has a command line interface that I can interact with. When I applied it to the new-format JSON files it warned that there was a problem with the schema, but I did not understand the warning and saw that it was flagged as a "strict mode" issue only anyway.

Keeping that in mind, I proceeded with my testing and found that it claimed the JSON was valid according to the schema. So I started to break the JSON in ways that I thought validation should pick up and found that it didn't. This looked interesting so I kept experimenting and cross-referencing with the JSON schema documentation until I discovered that part of the schema appeared to validate nothing and that it related to the strict mode warning in a way that now made some sense.

I could have reported a bug and finished at that point but I decided to try to fix the schema. Why? Because I already had a lot of context and I thought there was an opportunity to learn more. I made a handful of edits, pushed them back into the still-open PR and asked the developer who wrote the schema to check that I hadn't broken some assumption that he'd made.

Writing Definition Files

One of the products I work on is a platform that exposes services defined in reasonably complex files. It's a new product this year and so no-one has much experience of writing these files and, with understandable prioritisation, we haven't built tooling for creating them yet either. As the files are so fundamental for the platform I have taken every opportunity I can to write them which means I have picked up implementation tasks for demos, proofs of concept, and documentation.

Along the way I have built throwaway tools to help me to write and test the files and the exposed services, found issues in both the syntax and semantics of the files, and exploited the files as tools themselves to help me to test the platform.

There was no pressure to pick up these implementation tasks. I chose to do it because I thought there was advantage to me and the team if I did, and I pick up other implementation tasks on the same basis. I don't enjoy programming enough — and frankly I'm not good enough at it — to take on anything that makes large changes or requires deep language-level knowledge, but I will attempt bug fixes, or fiddle with the Jenkins pipeline, or sketch possible approaches in spikes.

So What?

So what has this got to do with Joep's axes? Well, nothing that challenges them significantly but I think that there are additional dimensions that Joep didn't represent that are also interesting, including:

• immediate result vs invest in the future
• fix the brokenness vs report the brokenness
• build the thing vs test the thing

I initially wondered if any of these might be covered by Joep's third axis, "improve the system." In the article, a footnote refers to this LinkedIn post from Maaret Pyhäjärvi which expands the idea but suggests that it is talking about the context in which work is done, rather than the system under test. I pinged Joep on Mastodon and he confirmed that was the case.

His original post does touch briefly on a related topic, that of roles and activities:

This confusion of people, roles, and process steps means we rarely talk about what we do in a smart way. In a more specific, more granular way. What are the different activities that need doing? What skills and perspectives are needed for those activities? Who on our team can contribute? 

It's a good point but I think there's another important aspect: why would any given person (or persons) do any particular task? Whose needs are being met, in what ways, to what ends? 

I believe that being aware of and intentional about those questions is important and I keep them in mind as I make, fix, and, of course and mostly, test.
Image: Bing Image Creator