Compiler and encoding

Revision as of 20:19, 30 May 2012 by Ted (Talk | contribs) (Validility)


From 6.7, the compiler has been equiped with a Unicode parser. The core of the parser only accepts UTF-8 source code, for simplicity and generality. Before source code is passed into the core process of parsing, it is preprocessed and converted into UTF-8.

Internals

Data Storage

Abstracted syntax tree now stores STRING_8 as UTF-8 data on each node. There are also different features exporting UTF-8, UTF-32 or the written bytes.

Here is an example of how a character é is represented at various levels.

Source encoding

UTF-8 (BOM)

ISO-8859-1

Bytes in source

0xC3A9

0xE9

1. {STRING_AS}.value

0xC3A9

0xC3A9

2. {STRING_AS}.binary_value

0xC3A9

0xE9

3. {STRING_AS}.value_32

0xE9

0xE9

4. {STRING_AS}.string_value_32

0xE9

0xE9

5. Runtime

0xC3A9 (STRING_8)

0xE9 (STRING_32)

0xE9 (STRING_8)

0xE9 (STRING_32)

Validility

Source code encoding is either explicitly or implicitly specified.

  • Explicit
    • File level: UTF-8 (BOM) is implemented
    • Class level: note clause (not implemented)
    • Configure file: .ecf (not implemented)
  • Implicit
    • Implicit encoding is taken as ISO-8859-1 for compatibility, if no source code encoding is specified.

The following table shows how manifest strings are validated by the compiler:

Explicit Encoding Implicit Encoding (ISO-8859-1)
STRING_8 manifest Unicode point (0-255)? Valid (taken as bytes)
STRING_32 manifest Valid Valid