Hi,
Some thoughts on this.
Is it worth fixing and releasing Avatica first then release the new Calcite
version that depends on it?
Looking around getSubject, is there alternative security calls we should be
making in its absence?
If the getSubject() is resulting in extra security (even if it is deprecated),
should we remove it only for java 23?
Kind regards, David
From: Ruben Q L <rube...@gmail.com>
Date: Thursday, 19 September 2024 at 19:46
To: dev@calcite.apache.org <dev@calcite.apache.org>
Subject: [EXTERNAL] Re: [DISCUSS] Towards Avatica 1.26.0
+1, I think the proposed plan seems reasonable.
Ruben
On Thu, Sep 19, 2024 at 5:31 PM Julian Hyde <jh...@apache.org> wrote:
> While working towards releasing Calcite 1.38.0, I noticed that JDK 23
> has been released, and decided to try to upgrade Calcite to support
> it[1]. And I found myself blocked by an Avatica bug, namely that
> Avatica uses a getSubject method that has been deprecated for some
> time and finally disabled in JDK 23. (Incidentally, Hadoop has the
> same problem [3]; people don't expect them to run on the current JDK,
> but we will need to disable our Spark and Pig adapter tests under JDK
> 23 until they fix it.)
>
> There's currently just one bug, to upgrade Avatica to JDK 23. This
> includes re-enabling the mode where we fail to compile if any
> deprecated methods are used, and removing uses of getSubject [2].
>
> I think we can release Calcite 1.38.0 with limited JDK 23 support
> (i.e. a few tests disabled), but we're going to need a new Avatica
> release in the next month or two, and it will need to fix the
> getSubject issue.
>
> What do you think?
>
> Julian
>
> [1] https://issues.apache.org/jira/browse/CALCITE-6587
> [2] https://issues.apache.org/jira/browse/CALCITE-6588
> [3] https://issues.apache.org/jira/browse/HADOOP-19212
>
Unless otherwise stated above:
IBM United Kingdom Limited
Registered in England and Wales with number 741598
Registered office: PO Box 41, North Harbour, Portsmouth, Hants. PO6 3AU
