Bug 53811

piece of code:

final TarOutputStream tarFile = new TarOutputStream(new FileOutputStream(pathName));
tarFile.setLongFileMode(TarOutputStream.LONGFILE_GNU);

....

TarEntry tarEntry = new TarEntry(path);
tarEntry.setSize(new File(pre_path+path).length());
tarFile.putNextEntry(tarEntry);
in = new FileInputStream(convertPath(choosedAudioPath)+path);
while((len = in.read(buf)) > 0)
 tarFile.write(buf, 0, len);

tarFile.closeEntry();
in.close();

(In reply to comment #1)
> piece of code:
> 
> final TarOutputStream tarFile = new TarOutputStream(new
> FileOutputStream(pathName));
> tarFile.setLongFileMode(TarOutputStream.LONGFILE_GNU);
> 
> ....
> 
> TarEntry tarEntry = new TarEntry(path);
> tarEntry.setSize(new File(pre_path+path).length());
> tarFile.putNextEntry(tarEntry);
> in = new FileInputStream(convertPath(choosedAudioPath)+path);
> while((len = in.read(buf)) > 0)
>  tarFile.write(buf, 0, len);
> 
> tarFile.closeEntry();
> in.close();

sorry, small mistake. it is:
in = new FileInputStream(pre_path+path);
instead of:
in = new FileInputStream(convertPath(choosedAudioPath)+path);

it stops with a non-ASCII names (Arabic file name) while setting -Dfile.encoding=UTF-8. does it support adding UTF8 names?

the same bug has been solved in:
http://jira.codehaus.org/browse/PLXCOMP-195

please check the patch applied there and consider apply the same to resolve this.

When a TAR file is being written, and:
1. One or more filenames are so long that they require GNU mode
2. The platform encoding is UTF-8
3. Any character in a long filename results in a multibyte UTF-8 sequence

then an exception is thrown with this text:
"Problem creating TAR: request to write 'xxx' bytes exceeds size in header of 'xxx' bytes"

This issue is been solved in 1.9 with the new TarOutputStream(FileOutputStream, "UTF8") constructor and setLongFileMode(TarOutputStream.LONGFILE_GNU);