[ruby-core:58921] [ruby-trunk - Bug #9171] [patch] use fstrings for symbol table

Issue #9171 has been updated by avit (Andrew Vit).


Would this allow the symbol table to be GC'd or is that a separate issue?
----------------------------------------
Bug #9171: [patch] use fstrings for symbol table
https://bugs.ruby-lang.org/issues/9171#change-43470

Author: tmm1 (Aman Gupta)
Status: Open
Priority: Normal
Assignee: nobu (Nobuyoshi Nakada)
Category: core
Target version: current: 2.1.0
ruby -v: ruby 2.1.0dev (2013-11-27 trunk 43880)
Backport: 1.9.3: UNKNOWN, 2.0.0: UNKNOWN


Here is a simple patch to use fstrings for the table backing symbols.

Unfortunately it causes a segfault in test/rdoc/test_rdoc_parser_ruby.rb. Maybe someone wants to investigate.

diff --git a/parse.y b/parse.y
index 8207ad7..b1f3112 100644
--- a/parse.y
+++ b/parse.y
@@ -10333,7 +10333,7 @@ register_symid(ID id, const char *name, long len, rb_encoding *enc)
 static ID
 register_symid_str(ID id, VALUE str)
 {
-    OBJ_FREEZE(str);
+    str = rb_fstring(str);
 
     if (RUBY_DTRACE_SYMBOL_CREATE_ENABLED()) {
 	RUBY_DTRACE_SYMBOL_CREATE(RSTRING_PTR(str), rb_sourcefile(), rb_sourceline());
diff --git a/string.c b/string.c
index 231bb2f..6ae33e3 100644
--- a/string.c
+++ b/string.c
@@ -138,6 +138,9 @@ rb_fstring(VALUE str)
     st_data_t fstr;
     Check_Type(str, T_STRING);
 
+    if (!frozen_strings)
+	frozen_strings = st_init_table(&fstring_hash_type);
+
     if (FL_TEST(str, RSTRING_FSTR))
 	return str;
 
@@ -8707,8 +8710,6 @@ Init_String(void)
 #undef rb_intern
 #define rb_intern(str) rb_intern_const(str)
 
-    frozen_strings = st_init_table(&fstring_hash_type);
-
     rb_cString  = rb_define_class("String", rb_cObject);
     rb_include_module(rb_cString, rb_mComparable);
     rb_define_alloc_func(rb_cString, empty_str_alloc);



-- 
http://bugs.ruby-lang.org/