DEV Community

Discussion on: Why committing straight to main/master must be allowed

 
dagnelies profile image
Arnaud Dagnelies

I guess I just experienced the pain of our recent enterprise-wide policy of creating branches for each tiny bug/feature associated with MRs/reviews. It simply adds beraucratic burden and leads to all the outlined problems. The conflicts are also not there initially, they occur due to delayed reviews. Aslo, sometimes, you must wait until something is merged before you can "build upon it". It just slows down everything and feels unnecessary most of the time. Like I said, for complex stuff or unfamiliar stuff, MRs and reviews definitely make sense and should be done. On the other hand, blindly enforcing for each tiny detail is IMHO just an unnecessary burden. Due to all the merging happening accross the board, the version history is also a lot less linear, and if someone merged it wrongly, more difficult to fix.

Thread Thread
 
190245 profile image
Dave

I manage a small team, and we look after almost 60 different components.

All of your problems can be solved, if there's a wider desire in the company to change those situations.

For what it's worth - I probably agree with you - we do a hybrid approach of "whatever makes the most sense for the situation we're in" - sometimes that's a PR, sometimes it's a hotfix, sometimes I'll just hop directly into the Production system to tweak something.

I suggest you speak with your development management, and with your PMO, along the lines of "hey, this recent policy feels like it will hurt delivery timelines, has X, Y and Z been considered already?"