+1 for running on separate branch before check in -----Original Message----- From: sebgoa [mailto:run...@gmail.com] Sent: Wednesday, December 11, 2013 1:12 AM To: Santhosh Edukulla; gir...@clogeny.com; dev@cloudstack.apache.org Cc: Prasanna Santhanam Subject: Marvin refactoring
Hi devs and Santosh and Girish, I am looking at Marvin lately and I am seeing lots of changes happening, especially: https://reviews.apache.org/users/santhoshe/ That's great to see effort for integration testing. However I am concerned that these changes (sometimes temporary fixes) are going straight in 4.2, 4.3 and master. How about creating a marvin refactoring branch and working there ? I believe that what is being done to Marvin is significant enough (renaming of files, change of logger structures, change of function names, addition of constants etc..) that it should be developed in its own feature branch and a merge should be called once the refactoring is done. Personally, I don't agree with some of the cosmetic changes: like move to camel case for function name (not pythonic even though pep8 compliant) or the addition of codes.py (different name maybe ?). Sometimes the api gets broken as well, like in cloudstackConnection. We should also have an open discussion about the import of https://github.com/vogxn/cloud-autodeploy in marvin/marvin/misc/build. That repo from prasanna while part of the CI infra is really custom config for Citrix internal infra (right ?). I don't think this has a place inside the Marvin code. When we build Marvin for instance, what happens to that code ? Does it get uploaded to pypi if we push Marvin to pypi ? Let me know what you think, thanks, -Sebastien
