Fork me on GitHub
#clojure-dev
<
2015-11-14
>
cfleming17:11:51

It’s a mess - I never thought Oracle would be able to manage Java worse than Apple.

cfleming17:11:58

But they’ve done a stellar job at it.

cfleming17:11:52

IntelliJ 15 comes with a bundled JDK8, but whether it works well is hardware dependent - I have to turn it off since it doesn’t work with my mid-2012 rMBP

cfleming17:11:12

Well, it does work, but is unstable and slow.

cfleming17:11:29

But the bundled JDK will never work with IntelliJ versions < 15

cfleming17:11:17

@ghadi: The only issue is that I can’t use a version of Clojure that relies on JDK > 6 for the code that runs Cursive itself.

cfleming17:11:44

The code for apps developed in IntelliJ/Cursive can use whatever version they like.

cfleming17:11:37

So if Clojure 1.9 drops JDK6 support I’ll be on JDK8 for a couple of years, I guess, and just add patches I need to my fork.

alexmiller21:11:33

To be clear, we have not even started discussing Clojure 1.9 yet, so there is no plan either way.

arrdem21:11:20

Does the same solution not cover both problems? Have an opt-in flag for indy support, off by default so that it doesn't give Colin or all N of the Android folks a hard time but server side users can opt into it for performance. Similarly have at least an all or nothing opt-in flag to enable static linking so that in the same spirit existing with-redefs code and compiler monkeying continues to work but users can opt into the static linking behavior if they accept its tradeoffs. I suppose you could argue that ^:static is opt-in and really what you want is a global opt-out rather than ^:static being a noop that can be "turned on" but either would work. But this feels like stating the obvious so maybe this has been discussed and discarded already. I can't tell since there doesn't seem to be a design page on the wiki or a feature tracking JIRA ticket, just the changelog entry.