We have recently activated Bitbucket and one of the dev. came up with these limitations. Anyone care on commenting these?
I spent some time with Pipeline and I think it is too limited and we will need to continue with CircleCI. Here the list of limitation:
Doesn't support starting a pipeline when a PR is made using a fork repository (this is our current way of working), this could be mitigated by using GITFlow instead of fork.
Pipeline deployment is limited to 10 environments and doesn't scale when we are using mono-repo since we have multiple deployments per repository, e.g. at the moment XYZ would require 15 environments. Composer/Front-end would have the same issue and probably worse.
Pipeline deployment variable cannot be shared across steps, so if you want to read a private key in more then 1 script you are doomed (or have to use some archiving hack)
Pipeline YAML anchor (shared scripts) don't support passing variables (e.g. calling the same YAML script with different parameters)
Thanks for your feedback!