The focus in M5 remains firmly on collections, as the centerpiece of the 2.13 release generally. M5 is our feature-freeze release for 2.13; from here forward, we’ll be closing existing open loops but not embarking on or accepting anything new.
To reiterate, we will only accept collection fixes and anything directly related to improving collection usage and the migration to 2.13 (e.g., improved type inference for typical usage of the new collections) RC1.
Any changes that do not directly impact this will have to wait for 2.13.1. The RC cycle needs a narrow focus, so that we can release 2.13.0 soon (and thus 2.13.1 ;-)). I’ve started moving these PRs to the 2.13.1 milestone.
Are performance improvements for collections still on the table for 2.13.0? I wanted to do these after my PR for SeqMap was merged since @retronym also did some indirect work to improve performance
Please note that I am on vacation until this Sunday August 12. I have already merged several changes to Scala.js due to new breaking changes in M5 wrt. M4, but several additional ones have made their way to master since then. I will need at least one week after being back from vacation to catch up with all those changes, then one more week to release a version of Scala.js that will be compatible with M5.
A quick update from the Scala team: We have cleared the ticket queue for M5 this week but there are still a few open PRs that we want to merge. We are therefore delaying the release of M5 until next week. Any new tickets and PRs should go to RC1 by default.
about 40 more PRs have been merged in the last 7 days
we’re down to three remaining open PRs on the M5 milestone, all blockers
all three are likely near resolution
there aren’t currently any additional blockers in other repos (scala/bug, scala/scala-dev, scala/collection-strawman)
after these three PRs are merged, before rolling M5 we’ll need to do a last round of QA via the Scala 2.13 community build, which currently has about 60 projects in it (compared to ~160 in the 2.12 build). Hopefully anything that turns up can be dealt with fairly quickly, knock on wood.
So right now we’re looking at Monday as the earliest possible build day.
Quick update: we’re now a bit torn on whether to include my proposed fix for https://github.com/scala/bug/issues/10762, as that regression has prevented running the test suites in quite a few projects. It was tricky to fix, so we don’t want to rush to merge https://github.com/scala/scala/pull/7119, but it also seems like it would help to have it in M5 so that we can more easily publish these libraries against M5 (without having to disable their test suite).
we’re leaning towards including https://github.com/scala/scala/pull/7119 and rebuilding early next week, as multiple projects in the community build are affected. if we do rebuild, we’ll also include a few VectorMap fixes