You are browsing a read-only backup copy of Wikitech. The primary site can be found at wikitech.wikimedia.org
User talk:Thcipriani/Deployments/Patch type criteria
Should we elaborate?
Reading through the points in the document, I thought it might not hurt to elaborate a bit on some of them. This document seems to do a good job covering which options are ok and which are not ok but doesn't go into detail about the nuance of the MAY points. Perhaps we should discuss which way is preferred in cases where it could go either way? The RFC definition of MAY doesn't feel like it applies to the way it is used here. In general the RFC doesn't feel like a perfect fit for the definition of these terms in the context of this document.
For example, consider adding a second level of bullets:
- "Any change to MediaWiki core, extensions, or skins MAY be deployed via the train"
- Although these changes MAY be deployed via the train, for non-trivial changes it is often more appropriate to deploy these changes via backports and that SHOULD be the default choice for all but the most trivial changes.
Motivation / justification / reason?
Would it be a good idea to add a few words of justification for why each point is the way it is? Something like:
- New extensions MUST be deployed via the train.
- Justification: they have a high likelihood of breaking sites for all users, and it's easier to roll back large changed done via the train
- Changes meeting the criteria for dedicated windows MUST NOT be deployed via the train and SHOULD be deployed via dedicated windows
- Justification: train conductors don't usually understand the aspect or subsystem being changed, and having people who do understand it in dedicated deployment windows means it's less likely to break things badly, and any breakages are more likely to even be noticed as breakages
I'm not happy with "Patch_type_criteria" Jeena suggested "Deployments/Per_patch_type" which is better.