[All Lists] [By Thread] [By Date] [Previous] [Next]
From: Devorah
Subject: Implementation requirements
Date: 12 Shevat 5785
I am beginning implementation of an identity block parser. The specification describes the format. I have questions about what a conforming parser must do.
Questions:
1. Tokenization is clear. Split on whitespace. But what constitutes a valid token? The specification says "any sequence of ASCII characters constitutes a valid token." Does this mean my parser should accept anything between whitespace boundaries?
2. The specification says implementations SHOULD accept unknown category codes without error. What should my parser return for unknown codes? A generic token object? A warning?
3. For known codes, should my parser decompose modifiers? If input is K+/-(++), should output include structured data for primary (+), secondary (-), and conditional (+++)?
4. What about malformed modifier syntax? K+++++ has five plus signs. Is this valid? Is K+- valid?
5. The specification permits single-token blocks. Should my parser accept a block with just BJ? Or flag it as potentially incomplete?
I can build any of these behaviors. I need to know which are required.
Thread: