| 24 Sep 2024 |
 Infinidoge 🏳️⚧️ | I don't think we should simply remove for the sake of removing, where possible, but when it's actively a maintenance burden it's worth it. | 19:38:55 |
 emily | sadly every package causes some kind of burden – if only by slowing down Hydra evaluation and build times and causing the channels to bump slower for everyone. but in this case it's a security burden too, since 22 is now EOL and someone has to go in and deal with every single copy of it and bump to 23 | 19:40:32 |
| emily | it seems like the main difference between Zulu and Temurin is that the former bundles OpenJFX. I don't know if it's possible for us to use our source OpenJFX build with Temurin or not? | 19:40:55 |
| Infinidoge 🏳️⚧️ | Hmm, that would be a problem | 19:41:30 |
| Infinidoge 🏳️⚧️ | OpenJFX is bundled in at compile time iirc | 19:41:38 |
| emily | ah, so we ship Zulu for macOS because it's the only option to get JavaFX there, I suppose | 19:43:33 |
| emily | I was told that nobody actually really cares much about JavaFX recently :) but maybe it would make more sense to standardize on Zulu instead of Temurin, then? | 19:43:56 |
| Infinidoge 🏳️⚧️ | Temurin is pretty much the standard binary release of OpenJDK, so I don't think we'll switch over | 19:47:10 |
| Infinidoge 🏳️⚧️ | (It's descended from the original AdoptOpenJDK, for reference) | 19:47:23 |
| Infinidoge 🏳️⚧️ | Plus not everything needs OpenJFX | 19:47:34 |
| Infinidoge 🏳️⚧️ | Ideally, we'd do a source build of JDK and JFX for Darwin, but that takes effort lol | 19:47:55 |
| emily | are there that many obstacles to a Darwin source build? | 19:48:52 |
| emily | OpenJDK doesn't have that many dependencies, does it? | 19:48:57 |
| emily | (I don't think there's too much reason to prefer one TCK-certified OpenJDK build over another if one meets our needs better though, since it's all the same stuff – but sticking with Temurin and leaving OpenJFX to source builds sounds reasonable to me) | 19:49:42 |
| emily | FWIW I use Darwin so I might be able to unblock source JDK, though I don't really want to spend too much time on it as I've only really ended up touching Java stuff because of my FFmpeg work or like now for security reasons :) | 19:52:43 |
| Infinidoge 🏳️⚧️ | My only connection to JDK has been Minecraft lmao
I just decided "Nobody else is doing this and I nerdsniped myself so this is my moment" | 20:21:25 |
| 25 Sep 2024 |
| emily | btw, it looks like the JDK 8 distributions will be ending support in ~2 years with the expected OpenJDK EOL | 03:20:11 |
| emily | the 6 years I was looking at was for Oracle builds | 03:20:23 |
| emily | but Temurin and Corretto are giving up sooner | 03:20:42 |
| Infinidoge 🏳️⚧️ | Wait | 03:20:44 |
| emily | (looks like Zulu is going for 6-8 more years though?) | 03:20:54 |
| Infinidoge 🏳️⚧️ | Oracle builds are dying after 6 years? | 03:20:54 |
| Infinidoge 🏳️⚧️ | I'll finally stop having to deal with java.com in 6 years? | 03:21:14 |
| Infinidoge 🏳️⚧️ | Yippie! | 03:21:16 |
| emily | https://endoflife.date/oracle-jdk lists 6 more years of extended support for JDK 8. but those are proprietary builds that aren't relevant to us | 03:21:23 |
| emily | I'm not really sure why there's so much disparity. the GPL OpenJDK licence means that any of these longer-term support builds (at least from not-Oracle) will have to release the source for their patches anyway | 03:22:12 |
| emily | but Temurin 8 going EOL will mean there won't be a bootstrap JDK for 8, so realistically we'll probably have to stop supporting it within a couple years | 03:22:37 |
| Infinidoge 🏳️⚧️ | Yep | 03:22:54 |
| Infinidoge 🏳️⚧️ | Minecraft players will suffer, probably | 03:23:07 |
| Infinidoge 🏳️⚧️ | Or more accurately will install lwjgl3ify | 03:23:33 |
