On 10/04/2010 05:45 PM, Benjamin Bentmann wrote:
> Hi Paul,
> 
>> Author: pgier
>> Date: Mon Oct  4 22:22:30 2010
>> New Revision: 1004461
>>
>> URL: http://svn.apache.org/viewvc?rev=1004461&view=rev
>> Log:
>> [MANTRUN-155] Set encoding for generated Ant build.  Patch from Anders
>> Hammar with some minor changes.
>>
>> Added:
>>      maven/plugins/trunk/maven-antrun-plugin/src/it/task-encoding-test/
>>     
>> maven/plugins/trunk/maven-antrun-plugin/src/it/task-encoding-test/pom.xml  
>> (with props)
>> Modified:
>>     
>> maven/plugins/trunk/maven-antrun-plugin/src/main/java/org/apache/maven/plugin/antrun/AntRunMojo.java
>>
>>
>> +        String encoding = project.getProperties().getProperty(
>> "project.build.sourceEncoding" );
>> +        if ( encoding == null )
>> +        {
>> +            encoding = DEFAULT_ANT_BUILD_ENCODING;
>> +        }
> 
> This approach violates the usual usage pattern for the source encoding
> as outlined in [0]. First, the direct query to the property cuts off the
> future migration to a POM element. Second, the default value for the
> property is not UTF-8 but the platform encoding.
> 
> Given that the build file for Ant is auto-generated and not meant for
> human consumption/editing, I don't see any reason why its encoding
> should be configurable by the user and be linked to the sourceEncoding
> property. As such, why not keept it simple and just always use UTF-8 for
> the encoding? Always using UTF-8 also ensures that the build file can be
> written at all, even if project.build.encoding is set to a charset that
> can't represent the characters given in the POM.
> 
> 
> Benjamin
> 
> 
> [0]
> http://docs.codehaus.org/display/MAVENUSER/POM+Element+for+Source+File+Encoding.
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
> For additional commands, e-mail: dev-h...@maven.apache.org
> 

Hi Benjamin,

My concern was that a character could be added to the POM that would not
be compatible with UTF-8, but I guess that is not an issue since it can
represent all the unicode chars.  I see your point that I shouldn't use
a property, it should have been a plugin param instead to maintain
forward compatibility if/when sourceEncoding becomes a POM element.

I change it to always use UTF-8 in r1005210.

Thanks!


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
For additional commands, e-mail: dev-h...@maven.apache.org

Reply via email to