[log in to unmask]" type="cite"> Hi Eduardo,

That's an excellent idea.  Such a trimming could produce language dependent fonts.

The reduced fonts should be based on actual usage.  The ASL Dictionary  in SignPuddles provides live statistics based on over 10,000 signs using the search algorithms of SignWriting Text.
http://signbank.org/signpuddle2.0/frequency.php?ui=1&sgn=4

Looking at the first  group of hand shapes:



You can process the data directly from the SPML files, or I can create an alternate form of the symbol frequency data.

We will also need one for Libras:
http://signbank.org/sp20/index.php?ui=1&sgn=46

I hope 2013 will be the year that the Libras Wikipedia project starts.  It's currently just a stub.
http://bzs.wikipedia.wmflabs.org

If you can get this working, I'll have to re-implement the original Unicode compromise proposal that I dropped in 2011.

We have FSW: formal SignWriting:
M518x529S14c20481x471S27106503x489 M518x533S1870a489x515S18701482x490S20500508x496S2e734500x468 S38800464x496

We have CSW: character SignWriting:
������������� ����������������������� �����

Here is ISW: intermixed SignWriting
M518x529���481x471���503x489 M518x533���489x515���482x490 ���508x496���500x468 ���464x496

Here is how the ISW could look after you create the smart font language specific reduction that only renders symbol glyphs.  The signs are still obscure, but a little more obvious.
M518x529481x471503x489 M518x533489x515482x490508x496���500x468 464x496

And once the characters and symbols are correctly rendered in graphite, we will be able to see the actual written sign.


Good luck,
-Steve