Fix VariantWriter overflow on 64-bit int

Integers in Godot are signed 64-bit ints (int64_t), but var2str used
int behind the scenes and would thus overflow after 2^31.

Also properly documented the actual bounds of int and the behaviour
when overflowing them.
This commit is contained in:
Rémi Verschelde 2019-02-21 11:21:41 +01:00
parent 16934c7411
commit b39e1df704
3 changed files with 23 additions and 8 deletions

View file

@ -1597,7 +1597,7 @@ Error VariantWriter::write(const Variant &p_variant, StoreStringFunc p_store_str
} break; } break;
case Variant::INT: { case Variant::INT: {
p_store_string_func(p_store_string_ud, itos(p_variant.operator int())); p_store_string_func(p_store_string_ud, itos(p_variant.operator int64_t()));
} break; } break;
case Variant::REAL: { case Variant::REAL: {

View file

@ -811,7 +811,7 @@
<description> <description>
Random range, any floating point value between [code]from[/code] and [code]to[/code]. Random range, any floating point value between [code]from[/code] and [code]to[/code].
[codeblock] [codeblock]
prints(rand_range(0, 1), rand_range(0, 1)) # prints 0.135591 0.405263 prints(rand_range(0, 1), rand_range(0, 1)) # prints e.g. 0.135591 0.405263
[/codeblock] [/codeblock]
</description> </description>
</method> </method>
@ -830,7 +830,7 @@
<description> <description>
Returns a random floating point value on the interval [code][0, 1][/code]. Returns a random floating point value on the interval [code][0, 1][/code].
[codeblock] [codeblock]
randf() # returns 0.375671 randf() # returns e.g. 0.375671
[/codeblock] [/codeblock]
</description> </description>
</method> </method>
@ -838,11 +838,12 @@
<return type="int"> <return type="int">
</return> </return>
<description> <description>
Returns a random 32 bit integer. Use remainder to obtain a random value in the interval [code][0, N][/code] (where N is smaller than 2^32 -1). Returns a random unsigned 32 bit integer. Use remainder to obtain a random value in the interval [code][0, N][/code] (where N is smaller than 2^32 -1).
[codeblock] [codeblock]
randi() % 20 # returns random number between 0 and 19 randi() # returns random integer between 0 and 2^32 - 1
randi() % 100 # returns random number between 0 and 99 randi() % 20 # returns random integer between 0 and 19
randi() % 100 + 1 # returns random number between 1 and 100 randi() % 100 # returns random integer between 0 and 99
randi() % 100 + 1 # returns random integer between 1 and 100
[/codeblock] [/codeblock]
</description> </description>
</method> </method>

View file

@ -4,7 +4,21 @@
Integer built-in type. Integer built-in type.
</brief_description> </brief_description>
<description> <description>
Integer built-in type. Signed 64-bit integer type.
It can take values in the interval [code][-2^63, 2^63 - 1][/code], i.e. [code][-9223372036854775808, 9223372036854775807][/code]. Exceeding those bounds will wrap around.
[code]int[/code] is a [Variant] type, and will thus be used when assigning an integer value to a [Variant]. It can also be enforced with the [code]: int[/code] type hint.
[codeblock]
var my_variant = 0 # int, value 0
my_variant += 4.2 # float, value 4.2
var my_int: int = 1 # int, value 1
my_int = 4.2 # int, value 4, the right value is implicitly cast to int
my_int = int("6.7") # int, value 6, the String is explicitly cast with [method int]
var max_int = 9223372036854775807
print(max_int) # 9223372036854775807, OK
max_int += 1
print(max_int) # -9223372036854775808, we overflowed and wrapped around
[/codeblock]
</description> </description>
<tutorials> <tutorials>
</tutorials> </tutorials>