E-Mail Routing and Autoresponder Rules and Latin Character Sets

[Gerry]

When processing email routing rules, auto responder routing rules and other areas where expressions are used we us an SQL-like expression language.  A typical expression might be subject LIKE 'PR00%' which would do what you would expect.  Our current implementation follows the SQL-92 recommendation to make the LIKE operator case insensitive so subject LIKE 'PR00%' and subject LIKE 'pr00%' would both match fine.  However, our implementation only followed the case-insensitive rule for US-ASCII, that is [A-Z] [a-z] which presented challenges for our customers who work in other languages. 

We have now implemented case-insensitive LIKE matching for all Latin character sets including Cyrillic, Greek, Coptic, Armenian and Georgian as per UCS-2 Level 7.3. While this does not expand the case-insensitive matches for languages that require Case Folding or Semantic Normalising, this does expand the usefulness of LIKE expressions considerably beyond the previous implementation of US-ASCII only case insensitive matching.

Here are some examples that will now properly match even though case is different

• 'Автоматический ответ' LIKE 'автома%' - Cyrillic/Russian
• 'À Á Â Ã Ä Å Æ' LIKE 'à á â ã ä å æ%' - Expanded Latin (all characters, just the A's shown here
• 'Α Β Γ Δ Ε Ζ Η Θ Ι Κ Λ Μ' LIKE 'α β γ δ ε ζ η θ ι κ λ μ' - Greek alphabet (only half of it shown)

These changes are currently in dev testing, they will be rolled out to beta by the end of this week and live the following week (subject to no issues being identified)

Gerry