Can I ask what specific pain points you have with Gitlab CI? I've been using it extensively the last couple of years and all in all it's been a pretty smooth experience!
* Doesn't scale. In many places. In particular, GitLab CI is required to use GitLab's Gitaly for Git replication. Gitaly has architectural limitations on throughput, which are surprisingly easy to hit with even a medium sized team. Certain otherwise reasonable patterns, particularly child pipelines, cause severe load amplification with Gitaly.
* Bad error handling. Ruby errors leak through. Although the underlying error is my user error, returning it to me as an underlying Ruby exception is really unhelpful. It basically shows that they didn't validate inputs at all. The input is trusted, so it's probably not a security issue, but it's a huge usability and developer experience issue.
* Config format is weird and deficient. Surprisingly difficult to programmatically generate. Can't generate new jobs into the current pipeline. Must generate child pipelines. See above for the load amplification issue therein.
