Letterbook has a smallish set of integration tests, and yesterday I wrote some assertions covering side effects after the response based on the traces they produce.
-
Letterbook has a smallish set of integration tests, and yesterday I wrote some assertions covering side effects after the response based on the traces they produce.
Which is amazing! It works great; took 30 minutes. It's exactly the kind of thing I've been trying to make happen at various jobs for 4 years. But I've always been prohibited, for mumble mumble feature velocity.
I feel like I should celebrate, but I also feel like no one will properly appreciate the vindication
go me, I guess ?
-
Anthony Steelereplied to Jenniferplusplus last edited by
@jenniferplusplus we have great feature velocity specifically because we lean heavily on outside in tests. Don't call them integration tests, though.
-
Jenniferplusplusreplied to Anthony Steele last edited by
@anthony_steele I still think it's important to be able to exercise and manipulate things in small units. But it's also important to exercise production-like compositions. As long as it's reliable and fast.
-
Jenniferplusplusreplied to Jenniferplusplus last edited by
@anthony_steele But I do strongly agree that being testable and observable does not detract from velocity, and actually increases it
-
Hrefna (DHC)replied to Jenniferplusplus last edited by
@jenniferplusplus That is amazing and worthy of celebration.
That is also something I have fought for, tooth and nail, across multiple jobs for years >_>
-
Jenniferplusplusreplied to Hrefna (DHC) last edited by
@hrefna it's so weird that "I can work better if I can actually tell what the software is doing" is something people fight so hard against
-
@[email protected] worth celebrating indeed … go you!
-
Anthony Steelereplied to Jenniferplusplus last edited by
@jenniferplusplus Of course, I too believe that "being testable and observable does not detract from velocity".
Though since the DORA people and the book "Accelerate" by Forsgren, Humble and Kim, 2018, did the data and the science on how quality and velocity *correlate* rather than being a trade-off, this is link not just a "belief" , it is an evidence-based view. i.e. a "to the best of our knowledge" scientific fact
-
Jenniferplusplusreplied to Jenniferplusplus last edited by
I blogged about collecting traces in my tests so that i can assert on them.
Test using OpenTelemetry traces in Asp.Net
Taking advantage of our OpenTelemetry tracing to easily test behavior that is otherwise very hard to observe
Jennifer++ (jenniferplusplus.com)
Doing these things is faster, easier, and more valuable than you think it is. Good telemetry is game changing.
-
@jenniferplusplus oh yeah. A large amount of the tests I wrote at my former employer used telemetry datapoints as the trigger, so our tests weren't spinning. Two birds, one stone. faster, more reliable tests, and free telemetry testing.
-
@ashteranic Maybe some day I'll be able to have these things at work
-
@jenniferplusplus It was pretty neat. We were using ETW (Event Tracing for Windows). Our telemetry, logging, tests and crash dump reporting could all consume it simultaneously. Sadly, Windows-specific, not sure I've run into a good lightweight similar solution that's platform agnostic (not that I've spent a lot of time looking yet)
-
@ashteranic @jenniferplusplus this is a great tip. I just started doing it in my tests.
-
Jenniferplusplusreplied to Roger Lipscombe last edited by