Hi all,
Our UAT teams have several hundred scripts which require specific data for each scenario. the scenarios or “test personas” can be reused in future changes and development and the associated data is continuously refreshed.
The suite of scripts are in place but not automated / batch job. The business periodically ask for test data to meet their tests and I would like to give them the ability to self-serve.
To start with I would like to get them a Report where they can complete the following:
- select which test data records they need using checkboxes.
- if they need to run the full suite (obvious performance issues!) they can do so with 1 click
- there is a description on the each scenario but maybe a potential to have further filters on common searches within the data.
- the data could be exported to excel for their use.
- the ability to define how many records (SELECT TOP 100) need to be returned in the resulting data.
- the ability by myself to continuously add and edit new scenarios within the test data report.
Screen mockup included.
Are there any OOTB or Template Reports on PowerBI / Tableau examples that would meet such a requirement? Maybe a web application needs to be developed?
Any feedback appreciated.
D
3 Likes
Hi Dave, I’ve never dealt directly with a UAT team before, is this subcontracted?
I suspect there is one big thing going worrying me here structure-wise, but then still a real need for a good tool… but that’s in itself another smoking gun. I’ve been planning to create such a tool myself @davecoleman , but not sure how it should look or work, you may notice a recent thread on this very same topic. Generic in-house test data generator with a GUI. What might this look like? - #12 by devtotest
Problem #1 is that “the business” are asking “the builders” for a template machine that creates data that is following business logic and is thus data-consistent, to rules which “the business” itself created. The builders of the software are probably not supposed to be separate from the business, but it can happen. So ideally, they should be configuring the tool whatever the tool is. This is a warning that you might have a “team” that is being lazy, by just asking for data. If not actually thinking about what we do, the minute we stop thinking, we are not longer testing anything at all, and the entire exercise stops being quality centric, since it’s not testing anymore: (James Bach words, not mine Why Scripted Testing is Not for Novices - Satisfice, Inc. . )
- I don’t disagree, you need a tool for this, but the tool needs to be configured by someone in “the business”, not by someone in “the builders” team.
Problem #2 Two teams, will want different things. Any such tool needs to be able to inject “bad user” data, or data that tries to break the business logic or the actual “input guards”. I seriously doubt that the first group of users of the tool want that at all, but your team clearly will want such a capability (I know I do.) I would also be wanting to regularly change and update the tool, for evil and also for good purposes (like differing app versions). This would be at variance with what the business want, and IMHO they are better of rolling their own generator app in the longer term as a way of creating a better responsibility as well as change management separation.
- Starting to think a lightweight “locally hosted” portable web app might do what we are wanting here, and have found a source of data which might be used - will need to check commercial license on doing that. My concern is that the definition of what fields to output and what “options” are valid becomes a DSL(domain-specific language) itself if we let it.
thanks @conrad.braam for the details. Have been away the past week (much needed break! 
)
the UAT teams are “Business” (non-technical) SME’s and as QA Manager I support those users to complete their testing. We aren’t a true software house.
The business in this instance wouldn’t have the technical ability to build a tool like this and hence why I’m exploring possible solutions.
The solution here would be very much a row by row selection by end users and a background SELECT query runs to find the data they require. The data is already obfuscated and is simply a persona they require for their specific test scenarios. there would be no data input / edit only by myself via a Query management tool if required.
I do believe it looks like I would need to involve our development team in building a portable application that can be easily updated.
2 Likes
For now I’ve implemented a test data SELECT framework via Python to connect to the database via JDBC.
Loosely it will
- Parse in an excel file with all the SELECT statements (over 500) which find TOP 5 records needed.
- Output the data to a CSV.
The SELECT will always output the Test reference number and the Customer ID required.
- Any timeouts mean a retry can occur.
- as there are over 500 scenarios, it will “cycle” through the excel.
