[
https://issues.apache.org/jira/browse/MSHARED-796?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16753670#comment-16753670
]
Hervé Boutemy edited comment on MSHARED-796 at 1/28/19 3:28 AM:
----------------------------------------------------------------
I’m not convinced by {{java.specification.vendor}}: I did not really
investigate which value it has, but either this value is the same whatever the
JDK distribution is, then it is not useful, either it’s different from
distribution to distribution then I don’t want it because it’s not reproducible
by “portable”, what do you mean?
I can assure you that {{java.specification.version}} is useful to reproduce a
Java build
was (Author: hboutemy):
I’m not convinced by {{java.specification.vendor}}: I did not really
investigate which value it has, but either this value is the same whatever the
JDK distribution is, then it is useful, either it’s different from distribution
to distribution then I don’t want it
by “portable”, what do you mean?
I can assure you that {{java.specification.version}} is useful to reproduce a
Java build
> use java.specification.version instead of java.version in Build-Jdk manifest
> entry
> ----------------------------------------------------------------------------------
>
> Key: MSHARED-796
> URL: https://issues.apache.org/jira/browse/MSHARED-796
> Project: Maven Shared Components
> Issue Type: Wish
> Components: maven-archiver
> Affects Versions: maven-archiver-3.3.0
> Reporter: Hervé Boutemy
> Assignee: Hervé Boutemy
> Priority: Major
> Fix For: maven-archiver-3.4.0
>
>
> as part or reproducible builds efforts, we're removing manifest entries that
> are hard to reproduce and do not add much value.
> In MSHARED-661, we removed "Built-By: <username>" entry, but decided to keep
> "Build-Jdk" entry
> the idea here is not to remove "Build-Jdk" entry but to replace its value
> from too much detailed java.version system property (which details the patch
> level) to java.specification.version: the specification has an impact on the
> generated bytecode, then is useful to keep, but at least we'll avoid patch
> level information
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)