Fluentd v1 now works on ruby 2.1 or later but it has several drawbacks.
dig method: See https://github.com/fluent/fluentd/pull/2372Removing 2.1/2.2 is good for the further development. td-agent and docker images use ruby 2.3 or later so it doesn't break popular environment.
However removing support without the transition period may break user environment. So support 2.1/2.2 until the end of 2019 is better.
UPDATE: Drop 2.3 together. See https://www.fluentd.org/blog/drop-schedule-announcement-in-2019
Thanks for opening this issue.
Maybe it's also better to discuss future support roadmap, I mean "when we drop support with Ruby2.3, 2.4, 2.5... for fluentd in the future" policy. Not fixed one but a rough consensus is good, IMO.
Maybe it's also better to discuss future support roadmap, I mean "when we drop support with Ruby2.3, 2.4, 2.5... for fluentd in the future"
One year support after EOL is good for me but several popular distributions use EOL version with own patches. So fixed drop timeline is not good for community.
Of course, we should care about distributions, and we can predict distros EOL thus suggest "plan" to the community, right?
distro | ruby version | date
------- | ---------------- | ------
Ubuntu 14.04 LTS | ruby 1.9.1 | EOL in this April
RHEL6 (CentOS6) | ruby 1.8.7 | November 2020
Ubuntu 16.04 LTS | ruby 2.3.1 | April 2021
Debian9 LTS | ruby 2.3.3 | June 2022
Ubuntu 18.04 LTS | ruby 2.5.1 | April 2023
Debian10 LTS | ruby 2.5.5 | May? 2024
RHEL7 (CentOS7) | ruby 2.0.0 | June 2024
RHEL8 (CentOS8) | ? | ? 2029
So, I suggest you
If current release supports 2.1 and so on, why can鈥檛 existing users continue to use it? Why do you need to support forward compatibility for LTS? Surely it鈥檚 enough to have a stable branch with security backports, etc. LTS indicates some desire for stability, so does it even make sense that users would want the latest release, or stick to a known good version?
Thanks for the update. I will probably remove 2.3 support too.
Nice work!