Originally Posted by rob
I have avoided going into too many details, since that's just airing dirty laundry, but I guess it's important for context on how we got here.
About 3.5 years ago, I nearly died, and the recovery was protracted. I was on truly nasty meds that both clouded my brain AND made it so I could only sleep about 3 hours out of every 24, which just compounded the brain fog. I talked regularly with the developer that messed up, and I provided what guidance I could in that mental state. This was a VERY senior developer, and I beseeched him to bring his "A" game, since I was seriously gimped. Alas, it turns out he was going through issues outside of work, and that massively impacted his work performance. So I wasn't able to provide the necessary oversight, and he brought his "D" game (the server limped with lots of issues). It wasn't malicious on his part, but it certainly was bad timing, and I was in no state to recognize the problems as they were unfolding.
After finally getting off the meds, it took me a full 18 months to finally flush them all out of my system. I began regaining a little clarity last spring (7-8 months after stopping the meds), and I finally was able to actually get a good night's sleep (read: 6 hours) this past April. Having some semblance of clarity last summer, it became apparent things were messed up with the server code, but I assumed it was a handful of issues that could be addressed and worked with the developer to address them. When that merely uncovered new issues, I launched a detailed code review and discovered the extent of the problems - they were fundamental design failures. At that point, all I could do was accept the reality of the situation and figure out how to salvage things, which is why the server code has now been rewritten.
That doesn't change the fact that we are now more than a year behind where we should be in terms of progress. And all the frustrations of users due to these setbacks is a reality we're slowing working to repair.
As for instituting things like pair programming, that's not really feasible with a development team of only 4, especially when spread across a range of dissimilar technologies (JS-client, C#, C++, etc.) and domains (client, server, HL engine, licensing/sales). We utilize peer-review of all checkins within the client, where we have two very capable developers with that expertise. However, we don't do that on the servers (C#) and in the HL engine (C++), where only a single developer has good domain knowledge right now.
We also employ peer-review of all checkins within the data file team, where we have a team of 6 working on those.
FWIW, we do use a CI/CD pipeline for the client-side development. And we have a dedicated QA person.
I'm not sure if that answers all the questions that people have, but that ought to at least provide some useful context.
|