This seemed like a good idea at the time but using the raw flags
directly is quite ok too, no ne...

Add some basic examples and bring some high level order to a document
that's grown organically.

The Julia flisp parser disallows things like `+₁ x` to mean
`(call +₁ x)`, so we probably should...

It seems we need to use peek_behind here, as a full expression of
arbitrary length (parsed by pa...

When reducing syntax disagreements, search for all failing subtrees
rather than the first one. A...

Also some tweaks to SourceFile pretty printing

Make peek() a method of Base.peek, as it's semantically similar.

Internally in the parser we'd like to work with a string of UTF-8
encoded source code, as we'd l...

Also replace the julia_version keyword with simply `version` in various
places and simplify the ...

In https://github.com/JuliaLang/julia/pull/39594 it was considered
non-breaking enough to change...

Work on defining a more useful public parser API in the combination of
- parse() and build_tre...

Also add a bunch of test cases for numeric literal parsing and fix the
types which arise from pa...

* Fix parsing of dotted short circuiting operators .&& and .||
* Fix parsing of dotted type comp...

Our parse_factor / parse_factor_with_initial_ex is a bit different from
the flisp parser so type...

* Add Unicode NFC normalization of all identifiers
* Add Julia-specific normalizations of lookal...

Introduce a category for contextural keywords. Use that to ensure that
strings are parsed as raw...

Removing this allows us to remove the VAR_IDENTIFIER/VarIdentifier token
kind, the VAR_EOF token...

It turns out that having this special token type in the lexer isn't
actually very natural. Inste...

This includes some simple tooling for test case reduction. This isn't
very precise yet and in va...

Split parse_assignment up so that we can reuse the second part -
parse_assignment_with_initial_e...

It turns out that the flisp parser does allow this, so we should too.

In `import A.==` the `.` is a separator, not part of a dotted operator
`.==`. Fix this. Also fix...

Now that we have a real parser, we can just use this parser for the
syntax interpolation prototy...

This seems to be strange syntax, and I guess unintentional that this was
allowed by the flisp pa...

Many small fixes to correctly parse Base (some failures remain in the
stdlib and Base tests).

*...

There's quite a wide variety of syntax allowed in the function "name"
component which makes dist...

ParseState is very specific to Julia parsing, whereas the ParseStream is
somewhat less so and mi...

This allows the bit size to be interpolated.

Not yet connected, as it causes a few failures still.

The previous version turned out to be quite broken - initial whitespace
is never regarded as ind...

It turns out that things like K"Keyword" are very easy to accidentally
misuse.

For example `@eval (x,y)`

It seems that SyntaxNode shouldn't simply be a proxy for Expr, but
rather a more convenient inte...

Start to clean things up to give a better overview.

Probably no point keeping this working at this early stage.

* String unescaping
* Raw string unescaping
* De-indentation of triple quoted strings

This allows any of the tree types to be conveniently produced using the
parse_all interface.

Emit strings as either STRING or CMD. This distinction is required
downstream from the lexer whe...

This macro makes sets of kinds easier to read. We could also consider
optimizing the set members...

* Fix parsing of where with curlies; add test cases.
* Don't emit blocks for right hand side of ...

This is done in the lexer rather than the parser as the next character
is readily available.

This change completely reworks how tokenization of strings happens so
that expressions within st...

This bug caused the output stream to get out of sync with the text, with
all sorts of messy cons...

* Make short-form SyntaxNode printing use bare Symbols so that tests are
much nicer to look at...

We need several significant changes to Tokenize, so bringing that
library in here for now.