This page has moved to ecsharp.net.

LLLPG FAQ

Q. What are the advantages of LLLPG over other tools?

Many! See the beginning of article #5.

Q. I’ve used other parser generators before. Could you explain quckly how to use LLLPG?

That’s covered in article #1. Skim it… I haven’t written a cheat sheet yet. You might even want to read article #5, which was written 1.6 years later, because it talks about the newest features and summarizes the main features again, as well as the most advanced topics.

Q. What example grammars are available?

Other than the various examples in the articles, you’ll find three larger examples in the release repo - click “Download ZIP” on the right side. Note that the included Enhanced C# parser may not be up-to-date, you’ll want to pull the Loyc repo for that.

Q. How do I handle keywords properly in my input language?

See the “Keyword parsing” section in article #5.

Q. How do I avoid memory allocations during parsing?

See “How to avoid memory allocation in a lexer” in article #5.

Q. How do I use LLLPG without a runtime library?

This is demonstrated by the “CalcExample-Standalone” example in the release repo.

Q. Tell me about the runtime library (LoycCore in NuGet). Why three assemblies? ANTLR only has one!

I found the .NET Framework never had all the facilities I needed, so I made a series of libraries to “fill in the gaps” in the .NET Framework. You can read about them on their home page. Basically,

Article #3 talks about these libraries in more detail; see “A brief overview of the Loyc libraries”.

Q. How do I customize error handling in my grammar?

Please see “Error handling mechanisms in LLLPG” in article #3.

Q. How do I shut off an ambiguity warning?

If you have two ambiguous alternatives like (A | B), change it to (A / B). The only effect is to shut off the warning. You can mix | and / in a single series of alternatives, and warnings are suppressed only if there is a continuous chain of slashes between two ambiguous alternatives. For example, in (A | B / C / D | E / F), an ambiguity warning between B and D would be suppressed, but a warning of ambiguity between D and F would not be suppressed.

If there is ambiguity with an exit branch ([...]*, (...)+ or [...]?), use greedy to tell LLLPG to prefer to stay in the loop (or case of an optional item, to match the item.) nongreedy also exists, but requires caution; for more information about these, see “Managing ambiguity, part 4: greedy and nongreedy” in article #4.

Q. What does this error message mean?

Could you be more specific? If you’re a beginner and wondering why something is ambiguous, make sure you understand LL(k) (see the next question).

LLLPG accepts “augmented” LL(k) grammars, which are the “top-down” grammars that decide which branch to take in the grammar before matching (consuming) anything from the input stream. By “augmented” I mean that LLLPG also supports extra features: zero-width assertions, “gates”, and mechanisms for ambiguity resolution (prioritization).

Please see “Parsing terminology” in article #2 for a more in-depth explanation.

Q. LLLPG generates a *.cs file from my *.ecs file. Should I check it into source control (Git/SVN)?

Yes. In fact, if you’re using the LLLPG Custom Tool in Visual Studio, LLLPG is not invoked when you build your project, so failing to commit it is a recipe for failing builds.

Q. How does LLLPG distinguish terminals (tokens) from nonterminals (rules)?

Some parser generators use uppercase and lowercase letters for this. Not LLLPG. Instead, LLLPG assumes that anything that is not a rule is a terminal. So if you refer to Foo in your grammar, it is a nonterminal if there is a rule called Foo, otherwise it is assumed to be a terminal.

Q. My question isn’t here!

I’m gonna level with you. This is not a real FAQ. In fact I’ve never been asked any of these questions. I am a fraud, I made them up. But you can still ask - reach me by email at gmail.com, with account name qwertie256.