[Pnet-developers] [bug] resgen produces invalid PO files

[Bruno Haible]

Using pnet-0.6.0.

============================ hello.de.resx ====================
<?xml version="1.0" encoding="utf-8" ?>
<root>
  <resheader name="ResMimeType">
    <value>text/microsoft-resx</value>
  </resheader>
  <resheader name="Version">
    <value>1.0.0.0</value>
  </resheader>
  <resheader name="Reader">
    <value>System.Resources.ResXResourceReader, System.Windows.Forms, 
Version=1.0.3102.0, Culture=neutral, PublicKeyToken=b77a5c561934e089</value>
  </resheader>
  <resheader name="Writer">
    <value>System.Resources.ResXResourceWriter, System.Windows.Forms, 
Version=1.0.3102.0, Culture=neutral, PublicKeyToken=b77a5c561934e089</value>
  </resheader>
  <data name="Hello World!">
    <value>Hallle Welt!</value>
  </data>
</root>
================================================================

$ resgen hello.de.resx de.po
$ cat de.po
msgid "Hello World!"
msgstr "Hall\xF6le Welt!"

The problem with this PO file is that it contains a non-ASCII character,
and therefore a header entry containing an encoding is needed in order
to interpret the byte sequence:

$ msgcat de.po
msgcat: input file `de.po' doesn't contain a header entry with a charset 
specification

The PO file format is specified in the GNU gettext documentation, section
'PO Files' and 'Header Entry'. A PO file with non-ASCII characters needs a
header entry, to specify the encoding of the PO file.