From 3f2dfe186bd3a203692a8832f3f29d13aa22498c Mon Sep 17 00:00:00 2001
From: Kaz Kylheku <kaz@kylheku.com>
Date: Mon, 7 Nov 2016 20:50:20 -0800
Subject: Support #: reading for uninterned symbols.

* parser.l (BTKEY, NTKEY): Renamed to BTKWUN and NTKWUN
("keyword and uninterned") respectively. Include an
optional match for the # character.
(BTOK, NTOK): Refer to BTKEY and NTKEY respectively

* parser.y (sym_helper): Implement uninterned symbols
by detecting when the package name string is "#"
and handling specially.

* txr.1: Documented package prefixes and uninterned
symbols.
---
 parser.l |  8 +++----
 parser.y | 16 ++++++++-----
 txr.1    | 78 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 92 insertions(+), 10 deletions(-)

diff --git a/parser.l b/parser.l
index 28e0d1d2..cff641ae 100644
--- a/parser.l
+++ b/parser.l
@@ -206,14 +206,14 @@ BT0     {BSCHR}({BSCHR}|{EXTRA})*
 BT1     @{BT0}+
 BT2     ({BSCHR}|{EXTRA})+
 BTREG   ({BT0}|{BT1})?:{BT2}?|({BT0}|{BT1})(:{BT2})?
-BTKEY   @?:{BT2}?
-BTOK    {BTREG}|{BTKEY}
+BTKWUN  @?#?:{BT2}?
+BTOK    {BTREG}|{BTKWUN}
 NT0     {NSCHR}({NSCHR}|{EXTRA})*
 NT1     @{NT0}+
 NT2     ({NSCHR}|{EXTRA})+
 NTREG   ({NT0}|{NT1})?:{NT2}?|({NT0}|{NT1})(:{NT2})?
-NTKEY   @?:{NT2}?
-NTOK    {NTREG}|{NTKEY}
+NTKWUN  @?#?:{NT2}?
+NTOK    {NTREG}|{NTKWUN}
 WS      [\t ]*
 REQWS   [\t ]+
 NL      (\n|\r|\r\n)
diff --git a/parser.y b/parser.y
index 48d36840..6a43f664 100644
--- a/parser.y
+++ b/parser.y
@@ -1249,7 +1249,7 @@ static val sym_helper(parser_t *parser, wchar_t *lexeme, val meta_allowed)
   int leading_at = *lexeme == L'@';
   wchar_t *tokfree = lexeme;
   wchar_t *colon = wcschr(lexeme, L':');
-  val sym_name = nil, pkg_name = nil, package = nil, sym;
+  val sym_name = nil, pkg_name = nil, package = user_package, sym;
 
   if (leading_at) {
     if (!meta_allowed) {
@@ -1270,13 +1270,17 @@ static val sym_helper(parser_t *parser, wchar_t *lexeme, val meta_allowed)
     free(tokfree);
   } else if (colon != 0) {
     pkg_name = string(lexeme);
-    package = find_package(pkg_name);
     sym_name = string(colon + 1);
     scrub_scanner(parser->scanner, SYMTOK, tokfree);
     free(tokfree);
-    if (!package) {
-      yyerrorf(scnr, lit("~a:~a: package ~a not found"), pkg_name, sym_name, pkg_name, nao);
-      return nil;
+    if (equal(pkg_name, lit("#"))) {
+      package = nil;
+    } else {
+      package = find_package(pkg_name);
+      if (!package) {
+        yyerrorf(scnr, lit("~a:~a: package ~a not found"), pkg_name, sym_name, pkg_name, nao);
+        return nil;
+      }
     }
   } else {
     sym_name = string(lexeme);
@@ -1284,7 +1288,7 @@ static val sym_helper(parser_t *parser, wchar_t *lexeme, val meta_allowed)
     free(tokfree);
   }
 
-  sym = intern(sym_name, package);
+  sym = package ? intern(sym_name, package) : make_sym(sym_name);
 
   return leading_at ? rl(list(var_s, sym, nao), num(parser->lineno)) : sym;
 }
diff --git a/txr.1 b/txr.1
index 6dd6285e..91127f58 100644
--- a/txr.1
+++ b/txr.1
@@ -10056,6 +10056,84 @@ variable syntax. Within \*(TL, regular expressions are written with
 a leading
 .codn # .
 
+.NP* Package Prefixes
+
+If a symbol name contains a colon, the
+.I lident
+characters, if any, before that colon constitute the package prefix.
+It is erroneous to read a symbol whose package doesn't exist.
+If the package exist, the symbol is interned in that package.
+
+If the package name is an empty identifier, the package is understood to be the
+.code keyword
+package and the symbol is a self-evaluating keyword symbol.
+
+For example
+.code foo:bar
+is the
+.code bar
+symbol in the
+.code foo
+package.
+
+The syntax
+.code :test
+denotes the symbol
+.code test
+in the
+.code keyword
+package, the same as
+.codn keyword:test .
+
+The syntax
+.code @foo:bar
+denotes the meta prefix
+.code @
+being applied to the
+.code foo:bar
+symbol, not to a symbol in the
+.code @foo
+package.
+
+The syntax
+.code #:bar
+denotes an uninterned symbol named
+.codn bar ,
+described in the next section.
+
+.TP* "Dialect note:"
+In ANSI Common Lisp, the
+.code foo:bar
+syntax does not create the symbol
+.code bar
+in the
+.code foo
+package; the symbol must exist or else the syntax is erroneous.
+In \*(TL, only the package has to exist; the symbol will be interned
+in that package.
+
+.NP* Uninterned Symbols
+
+Uninterned symbols are written with the
+.code #:
+prefix, followed by zero or more
+.I lident
+characters.
+When an uninterned symbol is read, a new, unique symbol is constructed,
+with the specified name. Even if two uninterned symbols have the same name,
+they are different objects. The
+.code make-sym
+and
+.code gensym
+functions produce uninterned symbols.
+
+"Uninterned" means "not entered into a package". Interning refers to a
+process which combines package lookup with symbol creation, which ensures
+that multiple occurrences of a symbol name in written syntax are all converted
+to the same object: the first occurrence creates the symbol and associates it
+with its name in a package. Subsequent occurrences do not create a new symbol,
+but retrieve the existing one.
+
 .NP* Consing Dot
 
 Unlike other major Lisp dialects, \*(TL allows a consing dot with no forms
-- 
cgit v1.2.3