Parsing System (v0.9)

FAQ

Q: How can I use Goldie to create a compiler?

A: This is not an easy task no matter what parsing tool you use. To get started, see the Never Used a Parsing Tool? page which contains an overview of everything involved in creating a full compiler.

In summary: Use Goldie to implement a lexer/parser (see Goldie Overview and the Beginner's Tutorial), and then, just as with any other parsing tool, you'll have to do your own semantic analysis on the resulting tree, and then implement a compiler backend.

Q: Why the name "Goldie"?

A: "GOLD for D" -> "GOLD D" -> "Goldie"

Q: How do I refer to a rule that has no subtokens?

A: Use null for the list of subtokens. For details and examples, see Referring to a rule that has no subtokens on the GoldieLib Overview page.

Q: Can Goldie parse the D language?

Short answer: Lexing: Mostly, yes (see the included dlex.grm). Parsing: Not yet.

Long answer: See this newsgroup discussion thread.

Q: I'm getting "Ambiguous symbol" errors!

A: See Ambiguous Symbols.

Q: This project sounds very similar to Gazelle. What's the difference, and which one is better?

A: GOLD/Goldie and Gazelle are both members of a new class of generic parsers that focus on fully-reusable grammars that aren't tied to any particular host language. They both accomplish this by completely separating the grammar from the parsing engine.

Both systems have their pros and cons, and which one's better is up to you. They're both evolving, so limitations on both sides are likely to change. Here is a comparison (disclaimer: based on my admittedly limited knowledge of Gazelle) so you can decide for yourself:

	Goldie	Gazelle
Parser Type	LALR(1)	LL(*)
Grammars Compile To	Tables	Bytecode
Written In	D	C
Easily Used From Other Languages	Yes	Yes
Event-Based Parsing	No	Yes
Creates a Parse Tree	Yes	No
Speed	Medium focus so far	Strong focus so far
Usability (Tools, API, Docs, etc...)	Strong focus so far	Medium focus so far
Existing Ecosystem	Fairly large, due to compatability with GOLD	Currently limited to the main Gazelle project
Grammar Definition Expressiveness	Powerful, but room for improvement	Appears to be quite powerful
Regular Expressions for Defining Terminals	GOLD's custom regex syntax, designed to be easy to read	Standard PCRE regex
Various Tools Available	Yes	Yes
Integrated Lexing and Parsing	Yes	Yes
Library API for Compiling Grammars	Yes	Yes
Strives for Flexibility	Yes	Yes
Cross-Platform	Yes	Yes
Commandline Compatible	Yes	Yes
Open Source	Yes, zlib/libpng	Yes, 3-Clause BSD

(If anything in this comparison appears to be wrong or out-of-date, I invite you to file a bug report.)

Q: What about Goldie vs. ANTLR?

A: While Goldie and Gazelle are in roughly the same class, ANTLR is very different in that grammars are not independent from the host language.

	Goldie	Gazelle	ANTLR
Parser Type	LALR(1)	LL(*)	LL(k)
Maturity	Still young	Still young	Very mature, with a large following
Official Host Languages	D	C	Java, C, C#, Python, ActionScript
Other Host Languages Available	Yes	No	Yes
Easy to Support a New Host Language	Yes	Yes	No. Typically, the entire ANTLR and StringTemplate projects must be ported.
Can Embed Source Code Into Grammar	No	No	Yes
Reuse Grammars on Other Host Languages	Yes	Yes	Typically, no. The embedded code must be ported.
Extensive Tools, Including GUIs	Yes	No	Yes
Open Source	Yes, zlib/libpng	Yes, 3-Clause BSD	Yes, 3-Clause BSD
Nutty Creator	Yes	Unknown	Yes

(If anything in this comparison appears to be wrong or out-of-date, I invite you to file a bug report.)

Q: Are these questions really frequently-asked?

A: No, I'm just anticipating most of them. ;)