ORD group wrong and incomplete

[GoogleCodeExporter]

If a linguistic problem:
What wordform makes the faulty analysis occur?
ORD group, located in NUM.fst is IMHO almost completely wrong. The problem is, 
that it gives back words, that might exist in the dictionary (or might not), 
and that mean something different, than the word to be analysed.
examples:
>ersten
erst<+ORD><Masc><Gen><Sg>
erst<+ORD><Masc><Akk><Sg>
erst<+ORD><Masc><Dat><Sg><Sw/Mix>
erst<+ORD><Neut><Gen><Sg>
erst<+ORD><Neut><Dat><Sg><Sw/Mix>
erst<+ORD><Fem><Gen><Sg><Sw/Mix>
erst<+ORD><Fem><Dat><Sg><Sw/Mix>
erst<+ORD><NoGend><Gen><Pl><Sw/Mix>
erst<+ORD><NoGend><Akk><Pl><Sw/Mix>
erst<+ORD><NoGend><Dat><Pl>
erst<+ORD><NoGend><Nom><Pl><Sw/Mix>

I would expect erste somewhere.

>zweiter
zweit<+ORD><Masc><Nom><Sg><St/Mix>
zweit<+ORD><Fem><Gen><Sg><St>
zweit<+ORD><Fem><Dat><Sg><St>
zweit<+ORD><NoGend><Gen><Pl><St>

I would expect zweite somewhere.

And so on

higher values are missing:
./fst_client tausendster
no result for tausendster

./fst_client millionster
no result for millionster

>hundertster
hundertst<+ORD><Masc><Nom><Sg><St/Mix>
hundertst<+ORD><Fem><Gen><Sg><St>
hundertst<+ORD><Fem><Dat><Sg><St>
hundertst<+ORD><NoGend><Gen><Pl><St>

This is the highest existing one, I would expect hundertste, as result.

I will try to fix this, and report the results.

Original issue reported on code.google.com by eleonor...@gmx.net on 25 Aug 2011 at 1:46

[GoogleCodeExporter]

I have tried to modify NUM.fst.
$Ord/3-999999$ = ( \
    $Ord/3-999$ | \
    (ein | $Card/2-999$)? tausendst |\    <---- added this; works
    (ein | $Card/2-999$)? tausend (und)? $Ord/1-999$ )

However, I am not able to fix the rest. :-(

Original comment by eleonor...@gmx.net on 25 Aug 2011 at 2:18