Has this been changed in the latest version of SOLR? I'm using v1.2.0 and it was unclear from the documentation whether it is valid to specify analyzers for non-text fields. I tried it when I wanted a simple tokenized string field, and it appears to work. Here's my field definition:
<!-- Fieldtype for simple tokenized strings. (Don't use for sorting, do use for space separated codes) -->
<fieldType name="stringText" class="solr.TextField" positionIncrementGap="0" omitNorms="true">
I use it for fields that have whitespace separated codes e.g. "101 104 15a 075". Just using a string field meant I could only find the exact full string rather than individual codes.