[ https://issues.jboss.org/browse/TEIID-5709?page=com.atlassian.jira.plugin... ] Steven Hawkins commented on TEIID-5709: --------------------------------------- This would require new functions: length2 and lengthc and probably a property to choose whether length means length2 or lengthc. Pushdowns would have to be adjusted accordingly. I'm leaning toward just documenting this instead. ascii - there is no consistency among sources as to the result when the first character is not in the ascii range. H2 provides the high surrogate value, pg provides the full code point value, oracle (not currently pushed) provides a bizarre value, sql server provides something forced into the ascii range. I'll just add a doc note about this behavior. initCap, translate, trim - I'll update the implementations to be be surrogate aware. I'll add a doc note about the 16 bit restriction. The code is straight-forward to update. -- This message was sent by Atlassian Jira (v7.12.1#712002)