RFC: backport of JDK-8210863: Remove Xrandr include files from JDK sources

[Andrew John Hughes]

On 30/07/2019 15:52, Aleksey Shipilev wrote:
> On 7/30/19 3:02 PM, Andrew John Hughes wrote:
>> What is the motivation for this? JDK-8213944 would seem to be evidence
>> that the removal risks breakage and OpenJDK 8u is an older JDK which may
>> still need to be built on systems without Xrandr.
> 
> Do we know such the system? Given this is already done in 11u, this would mean such a system could
> not build 11u already, and so would require updates.

I don't know every system on which OpenJDK 8 is being built. Given that
the change is to remove the header, rather than add it, the burden of
proof is on the person proposing the change to show it won't cause
breakage. The AIX issue, JDK-8213944, demonstrates that it is perfectly
possible for there to be systems already happily building 8u, only to
break when this header is removed.

I'm not sure what 11u has to do with anything. 8u is significantly older
than 11u and there will be older platforms where 8u is deployed, but 11u
never will be. Our own RHEL 6 is an example.

FWIW, I wouldn't have supported adding this to 11u either. Changes to
the build requirements should only be made to a legacy release when
really necessary. I see no such motivation here.

> 
> Also, the issue itself has "8-bp" label, which suggests Oracle would consider it for backports too.
> We can, of course, wait for that to happen.

I don't see how it makes any difference. This is one area where
comparisons with Oracle's trees don't make any sense. Oracle are
performing proprietary builds in a known build environment. If they have
these headers in all their build environments for JDK 8, they can drop
them from the source tree. As an open source project, we do not know all
the build environments used to build OpenJDK 8, so removing such a
header has a greater risk. Not only does this include a wider variety of
compilers and operating system versions, but it also includes a wider
range of architectures.

On the flipside, this is why Oracle don't tend to backport build changes
unless they are needed for their build environments, so backporting
changes to build with newer versions of GCC or work around a deprecated
readdir_r is pretty much left to the rest of us. They do make
alterations on occasion e.g. there was a flurry of build changes not
long before 8u changed ownership, to support building on a newer Windows
environment.

> 
> -Aleksey
> 

-- 
Andrew :)

Senior Free Java Software Engineer
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
https://keybase.io/gnu_andrew