Mihai, I'm in a similar position. I don't know (and don't wish to know) the details of how Gradle accomplishes things. However, Vladimir seems to believe this is the right solution, and seems to be OK with "owning this" if there are problems.
I know there are many fans of containerization (e.g. Docker) on this list. It seems to me that Gradle toolchains are a form of containerization and should be embraced. Julian On Wed, Sep 23, 2026 at 9:30 PM Mihai Budiu <[email protected]> wrote: > > I personally prefer to review changes that deal with the SQL compilation and > not the Java compilation of the compiler itself. That takes a lot of my time, > and I know more about that than about build tools. > > Tools are useful if they solve a problem. I don't understand exactly what the > benefit of these changes is for Calcite developers or users. > > Mihai > > ________________________________ > From: Vladimir Sitnikov <[email protected]> > Sent: Tuesday, September 22, 2026 7:12 AM > To: [email protected] <[email protected]> > Subject: Re: Replace the Checker Framework with NullAway and JSpecify > > Hi, > > Regarding "build with JDK 25 by default, fail if JDK 25 can't be > found/downloaded". > > From my point of view the silence looks terrifying as I cannot tell > "the change is too complex to review" from "looks ok" and from "let's > not do that because....". > > Hey, the change might impact everybody even though I hope it would be > invisible in practice. > I would like to gather more feedback rather than commit through the silence. > > Here's a PR of what I mean: https://github.com/apache/calcite/pull/5278 > It looks ok to me, and I leave it open for others to review or try in > their environment. > > PS. It's almost a month since I became a NullAway co-maintainer > PPS. There's a vote thread "accept NullAway into the Apache > Incubator": https://lists.apache.org/thread/f2l1v9j5j3qdrqy231sz44drwdzv94zb > > Vladimir
