# ACS Epic Retrospective
* once or twice a week check-ins are very helpful
* Underestimated size of ACS feature
* irregular syncups at the beginning
* PRs took days/weeks to get reviewed/merged
* Example: https://github.com/pulp/pulpcore/pull/1549
* Code that had already been reviewed once
* Took 2 weeks (Aug 9 -> Aug 25)
* Constantly nagged people for reviews/re-reviews
* People involved in the mini-team switched to another high-prio work for the 3.15 release, as a result ACS work was put on hold, PR/PR review left in a half-done state
* Multiple ptos/outages
* Possible solution: small PRs --> does not need half-day/day to load all on the code logic into the brain, easier to review, faster merges.
* Possible solution: feature branching
* 3-4 big PRs instead of 20-30 smaller ones
* Technical debt
* E.g. fixed task group responses
* Will want to factor in tech debt into future estimates
* Testing strategy and tests came at the end
* Would likely have been good to start with these up front
* Consciously deciding to stop or transition the work to others during outages would have been helpful
###### tags: `Minutes`