Sketch of ISO/IEC 30114-2 Information Technology -- Extensions of Office Open XML File Formats -- Character repertoire checking

[MURATA Makoto (FAMILY Given)]

I am wondering whether we should use an ignorable attribute or a child
of extLst for referencing CREPDL OPC parts from SML cells.

If we use an ignorable attribute, old implementations will throw it away.
(In my understanding, ignorable attributes will be thrown away by 
old implemetnations silently and this is the way our spec appears to 
mean.)

If we use a child of extLst, old implementations will not throw them
away.  But does this approach have its own disadvantages?

Cheers,
Makoto