[Translation-i18n] The robot doesn't understand my translation

[<go...@ud...>]

When I tried to send in some new translations to gcc 4.1.0, the robot
complains with a number of messages like these:

    gcc-4.1.0.sv.po:15949: 'msgstr' is not a valid GCC internal format =
string, unlike 'msgid'. Reason: In the directive number 1, the characte=
r '1' is not a valid conversion specifier.

The file contents looks like this:

    #: c-format.c:1758
    #, gcc-internal-format
    msgid "%s does not support the %qs %s length modifier"
    msgstr "%1$s st=F6djer inte %3$s-l=E4ngdmodifieraren %2$qs"

Apparently the robot validation doesn't understand the N$-reordering.
It is valid though.  Comments in the source says it is ok, and running
it confirms it:

    mimmi> env LANG=3DC gcc -Wall -pedantic -c c.c
    c.c:4:24: warning: use of C99 long long integer constant
    c.c: In function 'main':
    c.c:4: warning: ISO C90 does not support the 'll' printf length mod=
ifier
    mimmi> env LANG=3Dsv=5FSE gcc -Wall -pedantic -c c.c
    c.c:4:24: varning: anv=E4ndning av long long heltalskonstant enligt=
 C99
    c.c: I funktion "main":
    c.c:4: varning: ISO C90 st=F6djer inte printf-l=E4ngdmodifieraren "=
ll"

(I did validate on my own first, and got no complaints.  I guess that
is because I have a slightly older version of gettext, which does less
checking.)

What can we do about this=3F  Is there an updated gettext on its way
any time soon=3F