xgettext behaviour about i18n-complete HOT 1 CLOSED

Not a problem, proper documentation is still on the way anyway.
1 xgettext recommends that the code should always be in english, hence the translation key's encoding should not be a problem.
However, from xgettext --help you could use the --from-code=UTF8 and make sure you don't use --omit-header which in the current version 0.1.716.171241 seems to be in the Project/Properties/Localization/Localize.bat by default. I will remove it in the next version.

Also please make sure your .pot and .po files are saved as UTF-8 with or without signature.

Input file interpretation:
      --from-code=NAME        encoding of input files
                                (except for Python, Tcl, Glade)
By default the input files are assumed to be in ASCII.

     --omit-header           don't write header with `msgid ""' entry

2 For xgettext to correctly read attributes you need to use the __q flavor
i.e.

title=@__q("Gesamtfläche")

//would translate to 

title="Gesamtfläche" //with the translated text.

Please node that __q is similar with __ which accepts the message key as HTML

title=@__q("a < b")

//would translate into 
title="a &lt; b"

from i18n-complete.