biggest negative I would say is getting rid of the SDET role and moving them into Telemetry.
It moved the onus of testing on Devs who've never done it before as part of their job and thusly very lacking. The bungled W10 recent releases can be directly tied back to this model. As PMs and Devs don't account for testing like they used to before shipping to customers.
This may be fine and dandy short-term, but if you're biggest customers rely on stability for code bases over 25 years old--getting rid of focused testing and reliability is a big no-no.
> biggest negative I would say is getting rid of the SDET role and moving them into Telemetry.
I don't quite understand this move. Would you mind to elaborate?
Are you saying that previously SDET role exist at MSFT and today, it does not exist anymore? And that Developers have to rely on Telemetry (logging, metrics, notifications, error reporting, etc etc) ?
Perhaps I don't understand "Telemetry" in this statement.
They erased the Software Developer in Test role, and moved to a 'Quality' role.
The 'Quality' role implements Telemetry for the working product which turns into metrics that PMs/Dev Managers look at to determine shipping.
I used to be an SDET, and re-interviewed in order to be a Dev after the big switch. I could have rolled right into 'Quality', but I didn't find that as interesting. As a result, the automated Testing that used to be done by the SDET team is moved to be the responsibility of the Dev teams. Most dev teams knew little to nothing about the automated tests that were being run every day against their code (Upgrade tests, BVTs, integration tests).
SDETs also created tools and internal frameworks which were created to reduce cost; for example, I implemented a Code-Coverage based test-pass to reduce the 24 hour automated test suite that was running every day on Windows Phone main branch.
The previous SDET role's responsbility has moved 100% to Dev Teams, none of which had to ever think about that facet of development before. As a result, you get me implementing automated test suites for 5 different teams in 3 years (WSL, Ink, Text, Touch input, kernel sec) because I know the frameworks more than 25 year veterans of Windows internal development. Some of these teams had 10-15 years of built in collatoral that is now attempting to be revived because management realizes that they've dropped the ball quite a bit.
It moved the onus of testing on Devs who've never done it before as part of their job and thusly very lacking. The bungled W10 recent releases can be directly tied back to this model. As PMs and Devs don't account for testing like they used to before shipping to customers.
This may be fine and dandy short-term, but if you're biggest customers rely on stability for code bases over 25 years old--getting rid of focused testing and reliability is a big no-no.