On Wed, Dec 2, 2015 at 6:32 PM, Tim Starling &lt;tstarling(a)wikimedia.org&gt; wrote: Thanks for your caution here, Tim. A large merge requiring PHP 5.5 might be messy to unravel. Per our discussion earlier, I filed T120164 [1] proposing to institute a "last call" period for MediaWiki RfCs. As a result of this, we might not to use RFC meetings as the final decision on approval. I filed T120164 with a little hesitancy. I believe that we've made the RFC process more efficient these past few months, such that it should be possible to *reverse* an RFC quite simply: write another RFC. The more process we layer onto each individual RFC, the harder it is to use them for nimble decision making. We also discussed having some sort of minimum discussion time on wikitech-l prior to having the decision-oriented RFC meeting. That seems sensible, with the caveat that we discussed: we should ask that those proposing RFCs announce their intention on wikitech-l, directing the conversation toward the Phab ticket associated with the RFC. The Phab ticket should be the "discussion of record", whereas wikitech-l can be "announcement of record" (torturing the "newspaper of record" metaphor[2]) I think a "last call" period, if done correctly, can be a lightweight safeguard that can maximize participation and provide for welcome scrutiny. Done incorrectly, and it's a new way for stubborn defenders of the status quo to get everyone the hell off of their lawn. ;-) That's one of the things that worries me about making it too easy to reopening RFCs. It seems here in the U.S. we were able to repeal the 18th amendment with the 21st amendment. Even with a process as complicated as amending the U.S. constitution, the results were reversible. Our process (hopefully) isn't that complicated, but let's avoid making it more complicated by handwringing over the approval process. Rob [1] <https://phabricator.wikimedia.org/T120164> [2] https://en.wikipedia.org/wiki/Newspaper_of_record