damned lies, anjuta documentation and encoding bug

[Simos Xenitellis <simos lists googlemail com>]

Hi All,
When viewing the status pages such as
http://l10n.gnome.org/languages/zh_HK/gnome-2-24
one can see the error:

Download po file 
<http://l10n.gnome.org/POT/anjuta.HEAD/docs/help.HEAD.pot> anjuta 
<http://l10n.gnome.org/module/anjuta> [Error regenerating POT file for 
document help.HEAD: <pre>cd ] doSerialize(doc) File "/usr/bin/xml2po", 
line 602, in doSerialize outtxt += doSerialize(child) File 
"/usr/bin/xml2po", line 596, in doSerialize (starttag, content, endtag, 
translation) = processElementTag(node, repl, 1) File "/usr/bin/xml2po", 
line 496, in processElementTag myrepl.append(processElementTag(child, 
myrepl, 1)) File "/usr/bin/xml2po", line 496, in processElementTag 
myrepl.append(processElementTag(child, myrepl, 1)) File 
"/usr/bin/xml2po", line 496, in processElementTag 
myrepl.append(processElementTag(child, myrepl, 1)) File 
"/usr/bin/xml2po", line 496, in processElementTag 
myrepl.append(processElementTag(child, myrepl, 1)) File 
"/usr/bin/xml2po", line 534, in processElementTag translation = 
translation.replace(u'' % (i), replacement) UnicodeDecodeError: 'ascii' 
codec can't decode byte 0xe2 in position 95: ordinal not in range(128)" />

Source file:
http://l10n.gnome.org/POT/anjuta.HEAD/docs/help.HEAD.pot

It appears that the mere fact the POT file has a non-ascii character 
causes damned-lies to break.
The offending character apparently is … (ellipsis character), due to the 
0xe2 hint.
It is strange because other POT/PO files have non-ASCII characters as 
well, such as evince and epiphany.

If someone can figure out what's wrong, it would be great.

Talking about encodings in POT files, here is a recent discussion going on,
http://blogs.gnome.org/simos/2008/05/14/should-ui-strings-in-source-code-have-non-ascii-characters/

Simos