[jira] [Comment Edited] (GROOVY-9315) Bump bytecode version to 1.8

["Daniel Sun (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/GROOVY-9315?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16981217#comment-16981217 ] 

Daniel Sun edited comment on GROOVY-9315 at 11/25/19 8:23 AM:
--------------------------------------------------------------

bq. That place should be CompilerConfiguration.setTargetBytecode.
{{CompilerConfiguration#targetBytecode}} is something like 1.4, 1.8, which is can not be used in ASM API directly. So we have to map the {{targetBytecode}} to ASM's targetBytecode, e.g. {{Opcodes.V1_4}}, {{Opcodes.V1_8}}. The mapping logic is implemented in {{CompilerConfiguration.setTargetBytecode}} too:
{code:java}
    public void setTargetBytecode(String version) {
        setTargetBytecodeIfValid(version);
    }

    private void setTargetBytecodeIfValid(String version) {
        if (JDK_TO_BYTECODE_VERSION_MAP.containsKey(version)) {
            this.targetBytecode = version;
            this.asmTargetBytecode = JDK_TO_BYTECODE_VERSION_MAP.get(this.targetBytecode); // set ASM targetBytecode
        }
    }
{code}

As for {{asmTargetBytecode}}, it has only getter.

bq. What is the difference between "parse mode" and "compute mode"?
compute mode is for {{ClassWriter}}, e.g. {{ClassWriter.COMPUTE_FRAMES}}
and parse mode is for {{ClassReader}}, e.g. {{ClassReader.SKIP_FRAMES}}

bq. Can this wait for Groovy 3.0.1 or Groovy 4?
If you have more suggestions, we should have time to tweak the PR before Groovy 3.0.0-rc-2 release.



was (Author: daniel_sun):
bq. That place should be CompilerConfiguration.setTargetBytecode.
{{CompilerConfiguration#targetBytecode}} is something like 1.4, 1.8, which is can not be used in ASM API directly. So we have to map the {{targetBytecode}} to ASM's targetBytecode, e.g. {{Opcodes.V1_4}}, {{Opcodes.V1_8}}.

bq. What is the difference between "parse mode" and "compute mode"?
compute mode is for {{ClassWriter}}, e.g. {{ClassWriter.COMPUTE_FRAMES}}
and parse mode is for {{ClassReader}}, e.g. {{ClassReader.SKIP_FRAMES}}

bq. Can this wait for Groovy 3.0.1 or Groovy 4?
If you have more suggestions, we should have time to tweak the PR before Groovy 3.0.0-rc-2 release.


> Bump bytecode version to 1.8
> ----------------------------
>
>                 Key: GROOVY-9315
>                 URL: https://issues.apache.org/jira/browse/GROOVY-9315
>             Project: Groovy
>          Issue Type: Improvement
>            Reporter: Daniel Sun
>            Assignee: Daniel Sun
>            Priority: Major
>             Fix For: 3.0.0-rc-2
>
>          Time Spent: 2h
>  Remaining Estimate: 0h
>
> Groovy 3.0.0 requires Java 8+ and supports Java8 language features, so it's better to use its relevant class file version, i.e. 1.8
> In the old codebase, we can not manage the bytecode version well, for example, we use many bytecode versions to generate bytecode, some very old bytecode versions like {{1.3}} are still used.
> What I propose to do is manage the default bytecode code version in a place, same for its relevant compute mode and parse mode. If we want to change them, just modify only one place.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)