<github.com/cloudposse/atmos/pull/2898> ptal :grin...
# atmos-dev
z
e
Code rabbit has good suggestion
z
addressing 👍
I'm running this through another compound engineering skill before pushing to do a local review, I'm not sure if the ponytail skill is just fixing things too scoped or if coderabbit just hates me
e
"ponytail skill"?
Also, try the new
/field-test
skill ; it's AMAZING
z
e
I wouldn't say though field-test is lazy though; it's the most relentless QA test engineer.
z
atmos-emulator
ooo
e
This one is priceless
Makes a lot of things easier to test
z
Alright, I have beaten the rabbit into submission
Testing with this binary now and it looks correct
DAG with cross stack 🤌 feels good
Been following this thread. Can incorporate changes on Monday. Maybe there's skill in this repo that I should be using to steer my agent to follow specific standards? Your changes make sense to me. I'm not 100% if my last PR that fixed a bunch of the other DAG output followed the same patterns
👍 1
Maybe if there's a way to have coderabbit catch these types of things as well.
e
Yes, looks like we're missing a skill based on github.com/cloudposse/atmos/blob/…/io-and-ui-output.md
z
@Erik Osterman (Cloud Posse) Addressed with the skills
e
z
having the agent go through it
e
Then I'm satisified 😃
z
It's patching this 🤔
docs/prd/io-handling-strategy.md
e
That LGTM
z
Ok, this should be ready again
e
LGTM, kicked off the tests; if they pass, we'll merge
🙏 1
z
again pls, windows test failed 😭
oh, nvm, already going
idk how this thing works
@Erik Osterman (Cloud Posse) @Andriy Knysh (Cloud Posse) can I have more pipelines pls 🫠
🫡 1
idk if I'll ever pass windows acceptance tests
final boss
e
merged
z
shared a GIF
Idea - split the acceptance tests into a matrix of jobs?
Not sure if feasible
e
It's possible, and we probably should, just it's already split by 3. But with windows tests approaching an hour, it's annoying
z
Did you guys just implement merge queue also?
e
Yes, we're merge queueing now
That's a big help
z
Yeah, it's good. Unfortunately also doubles pipeline times because required status checks in both the PR and the queue
e
z
Did this end up speeding it up a lot?
windows test still taking 20 minutes a pop 😂
I guess cutting testing time down by 2/3 is good
a
not perfect, but much better than it was before
e
> windows test still taking 20 minutes a pop But its 10x jobs concurrently at < 20 minutes ea, instead of 1 job at 20*10
head shaking vertically 1