Implement printing in the specified radix for Number.prototype.toString(). #306
Conversation
dbatyai
added
ecma builtins
|
I've updated this to calculate an approximate buffer size. It can be slightly larger than necessary, but that should be negligible. I've added tests for two digit numbers as well. |
| } | ||
|
|
||
| uint64_t whole = (uint64_t) this_arg_number; | ||
| double fraction = this_arg_number - (double) whole; |
There was a problem hiding this comment.
Does usage of double instead of ecma_number_t improve precision in fp32 mode in the case?
There was a problem hiding this comment.
This was unintentional. But since you asked, I don't think this improves precision. The original value is already constrained to a float, and because of the continued multiplication the required precision to represent the result would only decrease.
There was a problem hiding this comment.
Ok, I see. The question was about additional precision lose, that may be can occur in intermediate operations.
I think we should use ecma_number_t in the case.
There was a problem hiding this comment.
Maybe, we should move the following defines from ecma-helpers-number.cpp to ecma-globals.h, near definition of ecma_number_t and use the definitions in the function:
#if CONFIG_ECMA_NUMBER_TYPE == CONFIG_ECMA_NUMBER_FLOAT32
JERRY_STATIC_ASSERT (sizeof (ecma_number_t) == sizeof (uint32_t));
/**
* Width of sign field
*
* See also:
* IEEE-754 2008, 3.6, Table 3.5
*/
#define ECMA_NUMBER_SIGN_WIDTH (1)
/**
* Width of biased exponent field
*
* See also:
* IEEE-754 2008, 3.6, Table 3.5
*/
#define ECMA_NUMBER_BIASED_EXP_WIDTH (8)
/**
* Width of fraction field
*
* See also:
* IEEE-754 2008, 3.6, Table 3.5
*/
#define ECMA_NUMBER_FRACTION_WIDTH (23)
#elif CONFIG_ECMA_NUMBER_TYPE == CONFIG_ECMA_NUMBER_FLOAT64
JERRY_STATIC_ASSERT (sizeof (ecma_number_t) == sizeof (uint64_t));
/**
* Width of sign field
*
* See also:
* IEEE-754 2008, 3.6, Table 3.5
*/
#define ECMA_NUMBER_SIGN_WIDTH (1)
/**
* Width of biased exponent field
*
* See also:
* IEEE-754 2008, 3.6, Table 3.5
*/
#define ECMA_NUMBER_BIASED_EXP_WIDTH (11)
/**
* Width of fraction field
*
* See also:
* IEEE-754 2008, 3.6, Table 3.5
*/
#define ECMA_NUMBER_FRACTION_WIDTH (52)
#endif /* CONFIG_ECMA_NUMBER_TYPE == CONFIG_ECMA_NUMBER_FLOAT64 */
5cfbfc1 to
7cd9c7a
Compare
dbatyai
force-pushed
the
number_prototype_tostring_radix
branch
3 times, most recently
from
74166b5 to
6e0e7a9
Compare
|
Looks good to me |
|
|
||
| if (should_round) | ||
| { | ||
| /* Round off last digit. */ |
|
+1 lgtm |
|
|
dbatyai
force-pushed
the
number_prototype_tostring_radix
branch
3 times, most recently
from
c9b907f to
af20005
Compare
…ng(). JerryScript-DCO-1.0-Signed-off-by: Dániel Bátyai dbatyai.u-szeged@partner.samsung.com
dbatyai
force-pushed
the
number_prototype_tostring_radix
branch
from
af20005 to
5e0a355
Compare
I had to re-open this, because I can't update the other one.
Previous discussion can be found here: #207
JerryScript-DCO-1.0-Signed-off-by: Dániel Bátyai dbatyai.u-szeged@partner.samsung.com