Hi Andrew, Chia-Ping Thanks for the suggestions. Personally, I also want to remove this in 4.0 since it is not a significant function. But as Chia-Ping mentioned, we seem to have a ‘1-year rule’ (or convention) for removing code, so I intend to follow the convention rather than being too aggressive. On the other hand, since this function doesn’t have much impact, maybe there’s a chance to ask Colin about his thoughts. What do you think?
Best Regards, TengYao Chia-Ping Tsai <chia7...@gmail.com> 於 2024年8月5日 週一 下午7:24寫道： > > So, I am in favour of the KIP, but I suggest that it should actually be > removed in 4.0 > > rather than deprecated in 4.0 and then removed in 5.0 because it already > doesn’t work. > > noted that removing it will introduce some broken behavior. For > example, the users who are using the config unintentionally will get > error "delete-config is not a recognized option" > > I'd like to delete it in 4.0, but we need to follow the depreciation > rules - remove it in major release and wait for 1-year > > There are some KIPs which don't wait for one year (see previous > discussion: > https://lists.apache.org/thread/bwqrsk341h0op662t1qhz781ld41lo5b), > but we agree they are exceptions. > > If this KIP-1079 can be considered to be another exception, it must be > included by 3.9.0. However, we should not merge KIP-1079 into 3.9 > unless Colin agrees. > > In short, the config is not critical and we try to follow a "complete" > deprecation cycle. Hence, it should be deprecated in 4.0 and be > removed in 5.0 > > Best, > Chia-Ping >
