8u232 Rampdown: Bug Triage

Tue Sep 3 09:53:53 UTC 2019

Hi,

On 9/3/19 6:30 AM, Andrew John Hughes wrote:

> We have reached the point of rampdown for 8u232 and need to decide which
> issues are still worth trying to push into that release and which should
> be deferred until 8u242.
> 
> I've been going through the list of bugs which are marked as fixed in
> Oracle's proprietary 8u231/2 [0] to see what is still missing in OpenJDK 8u:
> 
> 1. JDK-8222496/JDK-8198649
> "[8u] Switch on GTK3 as a default GTK L&F in client-libs"
> 
> This is 8222496 in 8u, but 8198649 in 11u just for extra confusion. I've
> see no review or approval request for this yet. As a simple default
> change, I'm happy for this to still go in early in rampdown while there
> is time for any side-effects to be tested for.

Seems a little risky at this late stage. I'd leave it until the next
cycle.

> It's a clean backport, so I've flagged it for approval.
> 
> 2. JDK-8219914
> Change the environment variable for Java Access Bridge logging to have a
> directory
> 
> This is in OpenJDK 13, but not 11. Not sure why. No sign of any review
> or approval for OpenJDK 8u as yet.

Punt 'til next time.

> 3. JDK-8219013
> Update Apache Santuario (XML Signature) to version 2.1.3
> 
> This is not yet in OpenJDK 8u or 11u. There are earlier Santuario
> changes it depends on.

Likewise.

> 4. JDK-8218629
> XML Digital Signature throws NAMESPACE_ERR exception on OpenJDK 11,
> works 8/9/10
> 
> No sign of review or approval. Again, depends on earlier Santuario
> changes, specifically JDK-8177334.

Likewise.

> 5. JDK-8218605
> Startup Splash Screen of SwingSet2 flashes in smaller coordinates before
> appearing in the final size
> 
> This was originally flagged for 8u222. There's no patch to backport and
> nothing has been posted for review.

Likewise.

> 6. JDK-8217878
> ENVELOPING XML signature no longer works in JDK 11
> 
> No sign of review or approval. Again, depends on earlier Santuario
> changes, specifically JDK-8177334.

Likewise.

> 7. JDK-8217581
> JDK 8 javadoc man page does not list correct values for -source
> 
> No patch to backport and no sign of a patch for review. Probably
> requires rather ugly changes directly to the man page as we don't have
> the sources checked in as far as I'm aware. Probably best to wait until
> Oracle update the online version.

Likewise.

> 8. JDK-8216401
> Allow "file:" URLs in Class-Path of local JARs
> 
> Patch has been posted, but needs to be updated following dependent
> ClassFileInstaller changes being backported (see [1])

This is a significant patch, and it should go in, but perhaps it's too
late for this cycle?

> 9. JDK-8215210
> [macos] Hangul text does not shape to the precomposed form on JDK8u
> 
> Similar to #5, this is an 8u222 era fix with no patch to backport and no
> patch posted for review.

Unless there's a MacOS engineer to step up, we can't do much about
this.

> 10. JDK-8205507
> jdk/javax/xml/crypto/dsig/GenerationTests.java timed out
> 
> Another that depends on JDK-8177334.

> 11. JDK-8196681
> Java Access Bridge logging and debug flags dynamically controlled
> 
> Similar to 8219914, but this one is in 11u at least. No visible attempts
> for 8u yet.
> 
> 12. JDK-8194653
> Deadlock involving FileSystems.getDefault and System.loadLibrary call
> 
> There was discussion on this before 8u222 in June 2019 [2], but no sign
> of any further consensus on a patch since.

It's not clear to me that this is a real problem in practice. I don't
know how frequently it occurs in ptactice, if ever.

> 13. JDK-8178870
> instrumentation.retransformClasses cause coredump
> 
> Patch has been posted but needs to be revised once JDK-8155951 is committed.
> 
> 14. JDK-8177334
> Update xmldsig implementation to Apache Santuario 2.1.1
> 
> Still in early stages of backport, so will need to defer this to 8u242.

I agree.

8177334 looks sugnificant, but I don't know why. All we have is "We
should upgrade this to the latest stable version in JDK 11."

> 15. JDK-8217676
> Upgrade libpng to 1.6.37
> 
> This needs some minor work still, but should be able to be pushed to
> 8u232 during rampdown.

OK.

> 16. JDK-8228469
> (tz) Upgrade time-zone data to tzdata2019b
> 
> This is a late arrival that isn't in 8u or 11u, but we should be able to
> get it into both during rampdown.

OK.

> 17. JDK-8229868
> Update Apache Santuario TPRM version
> 
> This should be combined with JDK-8219013 (#3)
> 
> 18. JDK-8222108
> 
> Reduce minRefreshTime for updating remote printer list on Windows
> 
> As with JDK-8217676, the patch needs some minor changes but should be
> possible to include during rampdown.

This doesn't look to be very important. Please consider if it's worth
doing in the critical phase.

> My current thinking is:
> 
> * Suitable for rampdown
>   - #1, #7, #8, #13, #15, #16, #18
> 
> * Defer to 8u242
>   - Access bridge patches (#2, #11)
>   - Apache Santuario work (#3, #4, #6, #10, #14, #17)
>   - AWT bugs (#5, #9)
>   - Deadlock bug (#12)
> 
> Other bugs may be considered during rampdown, but they should be
> regression fixes or minor bugs / test issues. Feature work should now be
> focused on 8u242.

I agree.

> I think things look better than the last cycle, but we are still seeing
> too many patches towards the end of the development period. If you want
> to target something for 8u242, now is the time to start working on it.
> For 8u232, the primary focus should be testing to make sure there are no
> major regressions.

I agree with that too.

Thank you for this very clear summary. It must have been a significant
effort.

-- 
Andrew Haley  (he/him)
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
https://keybase.io/andrewhaley
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671