This page is not created by, affiliated with, or supported by Slack Technologies, Inc.
To devs and dev leads here: which parts seem unrealistic to you? I am in amazement and disbelief at their efficiency working with this tremendously coupled legacy system. Setting up CI seems too easy for example. Or creating a shared dockerized environment for a component with so many dependencies. It is described as very messy yet still they’re able to make such huge improvements to the system in days/weeks, while in reality it looks more like months/years to me. Thoughts?
To devs and dev leads here: which parts seem unrealistic to you? I am in amazement and disbelief at their efficiency working with this tremendously coupled legacy system. Setting up CI seems too easy for example. Or creating a shared dockerized environment for a component with so many dependencies. It is described as very messy yet still they’re able to make such huge improvements to the system in days/weeks, while in reality it looks more like months/years to me. Thoughts?
It's not acutally a technical challenge: in my experience you just need one person to write the first test, automate the first deployment, commit the code to an SCM, … In the situation of TUP, Maxine takes the same approach I would take: start with one test, put it in CI and have the code automatically be tested, build, packaged, signed so the created artifact is ready to be deployed. Once you've got that part done, All you need to automate is getting that artifact package onto your servers and voila: you now have a CI/CD in place. One test leads to two, four, 10, 100, 1000 tests and as you progress, the quality gets better, the code more stable and your automation more fine-tuned. It's not rocket science, it's bravery to take that first step.
Maxine (nor anyone else in the organisation) had a clear view of all dependencies. Luckily Brent was there to help Maxine with her notes to get an environment for the developers to work with. Again, taking that first step was important.