[All Lists] [By Thread] [By Date] [Previous] [Next]
From: R. Feldman
Subject: Implementation requirements
Date: 14 Shevat 5785
The chair would summarize the discussion as follows.
There appears to be agreement that:
1. Tokenization (splitting on whitespace) is required. 2. Decomposition of known modifier syntax (+, -, >, /, ^, etc.) is parsing, not interpretation. 3. Recognition of specification-defined codes (K, O, BJ, etc.) is parsing, not interpretation. 4. Assertions about what the codes mean about the person would be interpretation and is prohibited.
The question of parsing levels is an implementation choice. The specification need not mandate a single approach.
I would note Devorah's practical concern: applications that wish to display structured information benefit from parsers that provide structure. This seems reasonable.
Regarding malformed syntax: the specification says implementations SHOULD accept unknown codes gracefully. I would suggest the same principle applies to unexpected modifier patterns. Flag them, do not reject them.
Does this summary reflect the committee's understanding?
Thread: