Backport LazyList

[NthPortal]

Still TODO:

• Fix compilation on 2.11
• Fix serialization

[NthPortal]

I wish I could mark this as a draft again, cuz boy is it not done.
help welcome

[julienrf]

@NthPortal Could you explain what remains to be done?

[NthPortal]

@julienrf port GC tests, and make all the tests pass in all versions (or comment out/fix those that can't due to compiler bugs that weren't fixed until 2.13)

[NthPortal]

if someone can take a look at the last two little bits and make them compile, that would be hugely appreciated (TODO list in description)

[NthPortal]

I've given up on getting the serialization test to work across versions. it will just have to be specific to certain versions (2.11 and 2.13 JVM)

[NthPortal]

it seems GC tests don't work on JS (idk why I missed this before). Is there a good way to let them run only on JVM? or can they be tweaked to use something other than Thread.sleep?

[NthPortal]

The docs for System.gc() say that when the method returns, the GC will already have been run (to the extent the "best effort" could), so it's not clear to me that the Thread.sleep(10) is necessary at all.

However, even without the Thread.sleep, the tests still fail for scala.js (probably because System.gc() is a NOP in scala.js). While JS engines do tend to have garbage collectors, there doesn't appear to be a way to manually invoke them (which is fair, from a browser design perspective).

I'm not sure what the best path forward is here, but if someone has one, they can feel free to (force) push it to this branch.

[yannbolliger]

The fix for this is here: NthPortal#1

[julienrf]

Thank you both!

[nilskp]

Do we know when this will be released?

[julienrf]

Hey, we should indeed cut a release soon, thanks for asking 😃
I think it would be good to have Lukas’ work in it. Let’s wait for a couple of weeks to see if we manage to merge it.

[lrytz]

Is there a reason not to release often? It's little work, we have everything automated, no?

[julienrf]

I think last time we had an issue and had to manually publish things. But I agree that in principle we could release more often.