RFR (M): 8062808: Turn on the -Wreturn-type warning

[Hohensee, Paul]

I didn't do a Zero build. But,

1. These patches went into 8u282, which is early in its development cycle, so there's a lot of time to shake them down. To me, pushing early == taking our time.

2. 8062808 was backported to 7u in 2018 with no issues, and platform support in 8u is essentially the same as in 7u. See https://www.oracle.com/java/technologies/jdk-jre-7-cs-config.html, https://www.oracle.com/java/technologies/javase/products-doc-jdk8-jre8-certconfig.html. So, I deemed 8062808 low risk, since it's been baking in 7u for a long time.

3. In the thread https://mail.openjdk.java.net/pipermail/jdk8u-dev/2020-February/011088.html, and this post from it https://mail.openjdk.java.net/pipermail/jdk8u-dev/2020-February/011237.html, you recommended as a pre-req for 8062808 the sequence 8022263->8030350->8036122. 8022263 was done in 8u252. 8036122 fixes an issue in 8030350, so those two were best pushed together. 8062808 could indeed have been pushed separately, but to me it seemed part of the sequence.

4. All three of 8030350, 8036122, and 8062808 are backports from 9, for which platform support was eventually restricted to 64-bit, see https://www.oracle.com/java/technologies/jdk9certconfig.html. But, 8030350 and 8036122 were pushed in early 2014, close enough to the end of 8 development (8 GA was March 2014) that 7's supported platforms were still being built. Given that, I deemed 8030350 and 8036122 to have low risk.

Thanks,
Paul

On 10/7/20, 10:00 PM, "Andrew Hughes" <gnu.andrew at redhat.com> wrote:

    On 20:15 Thu 01 Oct     , Hohensee, Paul wrote:
    > Reviving from https://mail.openjdk.java.net/pipermail/jdk8u-dev/2020-February/011088.html.
    > 
    > Xin’s last comment on https://bugs.openjdk.java.net/browse/JDK-8062808 notes that there are two remaining predecessor backports, https://bugs.openjdk.java.net/browse/JDK-8030350 and https://bugs.openjdk.java.net/browse/JDK-8036122, so this review request is for all three of them. They will be pushed together (but as separate changesets).
    >

    What is the motivation for backporting these two other fixes? I don't
    see any obvious reason to include these and I didn't when I backported
    JDK-8062808 myself.  Adding additional compiler warnings seems to be
    asking for build breakage.

    Though also not mentioned here, there is a motivation to backport
    -Wreturn-type as it catches cases which can lead to crashes with some
    compilers. I'm fine with that.

    If we must do all three, it would have made sense to at least put them
    in separate build promotions rather than rushing all three in together
    like this. I see absolutely no reason for that.

    > 
    > Compiles with both fastdebug and slowdebug on linux. I’d greatly appreciate if someone could try osx and solaris.
    >

    On what architectures? Have you checked Zero?

    > Thanks,
    > Paul
    > 
    > 
    > 
    > 
    > 

    Thanks,
    -- 
    Andrew :)

    Senior Free Java Software Engineer
    OpenJDK Package Owner
    Red Hat, Inc. (http://www.redhat.com)

    PGP Key: ed25519/0xCFDA0F9B35964222 (hkp://keys.gnupg.net)
    Fingerprint = 5132 579D D154 0ED2 3E04  C5A0 CFDA 0F9B 3596 4222