RFR: 8232365: Implementation for JEP 363: Remove the Concurrent Mark Sweep (CMS) Garbage Collector
Magnus Ihse Bursie
magnus.ihse.bursie at oracle.com
Tue Oct 22 09:42:20 UTC 2019
On 2019-10-21 17:45, Erik Joelsson wrote:
> On 2019-10-21 02:46, Magnus Ihse Bursie wrote:
>> On 2019-10-21 09:43, Per Liden wrote:
>>> Hi Erik,
>>>
>>> On 10/18/19 2:54 PM, Erik Joelsson wrote:
>>>> Hello Leo,
>>>>
>>>> When removing a JVM feature from the VALID_JVM_FEATURES list, it
>>>> needs to be added to the DEPRECATED_JVM_FEATURES list. Deprecated
>>>> here means removed here, but configure will still accept the
>>>> argument (with a warning).
>>>
>>> Are we really treating the list of JVM feature as a public/stable
>>> API, or is there some other reason for this? Would the plan be to
>>> remove it from the DEPRECATED_JVM_FEATURES in JDK 15?
>>
>> Note that this is an issue for building only. A feature listed as
>> deprecated in the build will not propagate outside the build system.
>>
>> We are basically treating the configure command line arguments as a
>> "stable API". The intention is that a command line on a build script
>> should not cause a build failure due to an unknown argument. We have
>> traditionally had 2-3 major releases before removing deprecated
>> command line options.
>>
> We (or at least I) have been a bit lazy with removing the deprecated
> args. I wouldn't mind being a bit more aggressive.
Be my guest. :-)
As long as it stays for at least one major release (even with the new 6
month release cadence), I'm ok with removing deprecated arguments. Otoh,
there's very little impact in the current deprecation system, so it does
not really matter if it lingers a bit more.
/Magnus
>
> /Erik
>
>
More information about the build-dev
mailing list