Changes:
- Add options to the test runner to be able to update exclude list
- Update the exclude list
- Make Travis CI signal if the exclude list isn't up-to-date
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
The output of the test262 runner changed between ES5.1 and ES6,
our script should respect it to signal regressions in return code too.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Ubuntu 14.04 reached its end of life on April 30m 2019.
Let's bump the reference to the latest LTS, which is 18.04.
Ubuntu 18.04 has newer Pylint and Cppcheck, the necessary
fixes and suppresses are also included in this PR.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
First step to fix issue #3568.
Changes:
- Implemented Math.clz32(), Math.fround(), Math.imul(), Math.hypot().
- Implemented all remaining Math functions with calling the standard
libm functions, but they throw UNIMPLEMENTED exception with jerry-libm,
because the necessary fdlibm functions are missing and should be ported.
All Math related test262 tests pass (except function name and length tests)
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Changes:
- Typo fixed in cmake build system and appveyor config
- Added a new buildoption test
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Changes:
- Crashing jerry / jerry-snapshot is always "FAIL",
only the 1 return code is accepted as expected fail.
- "FAIL (XPASS)" string is only used if the test marked
as expected fail, but the return code is 0.
- Simple "FAIL" string is used if the test marked as expected
fail, but the return code is not 0 or 1. (crash)
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Don't throw TypeError when object length is 0 and initialValue is undefined.
Additionally the error message should be more precise and match other engines.
Fixes#3463
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
The ecma_number_to_uint32 function is called many times, we can merge
ecma_number_is_nan and ecma_number_is_infinity checks to !ecma_number_is_finite
to let the compiler generate more optimal code for it.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Math.min() and Math.max() functions should apply ToNumber on each arguments
always, even if a NaN argument occured previously and the result will be NaN.
ECMA262 v5.1 15.8.2:
"Each of the following Math object functions applies the ToNumber abstract
operator to each of its arguments (in left-to-right order if there is more
than one) and then performs a computation on the resulting Number value(s).
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
There is no reason to force the developer to call ecma_number_is_nan
check before all ecma_number_is_zero and ecma_number_is_infinity.
These functions work fine and return false if NaN is passed.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Math.round(x) should be x it x is already integer. (ES5 15.8.2.15)
The current implementation calculates with x +/- 0.5, which isn't
representable numbers if x >= 2^52.
A correct bug fix can be to follow the spec and simply
return the number itself if it is already integer.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Math.trunc(x) should be -0.0 if -1 < x < 0 (ES2015 20.2.2.35). The problem
was that -0 isn't -0.0, but +0.0 in C. There were a test case for it, but
it was incorrect, because +0.0 === -0.0 in JS.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
If ecma_date_parse_year got an invalid date string, it could overread the input string.
The problem was that we compared the original str_p to str_end_p instead of str_start_p.
Additionally I simplified the parser loop.
Fixes#3388.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Code cleanup after #3314. Reverting already parsed '-' sign
after parse fail is unnecessary, because in this case the whole
date parse is failed.
Additionally this dead code was incorrect too, because it didn't
modify the original pointer to the string, but a local variable.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
REPL could read JS file from standard input when we passed "-" option,
but after #1896 "-" became invalid option mistakenly.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Changes:
- Parse output of Date.prototype.toString() and Date.prototype.toUTCString()
- Date.prototype.toString() is ECMA-262 v9 conform now, only TZ part changed
(Before ECMA-262 v9 it was implementation-dependent.)
- Reused day_names_p and month_names_p arrays (and made them more efficient)
- Tests updated and new tests added
Fixes#2946.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
JERRY_ATTR_NOINLINE define is __declspec(noinline) which should be placed
at the beginning of the declaration and caused build fail because it was
placed after the * character. Details can be found here:
https://docs.microsoft.com/en-us/cpp/cpp/declspec?view=vs-2019
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Changes:
- Use list comprehension instead of filter in get_tests
to ensure returning list instead of iterable object
- Fix unicode decoding issues with Popen call
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Uploaded JerryScript binaries aren't useful in practice
but this step regularly fails with an error message:
"Error uploading artifact the storage: Unable to connect to the remote server"
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Timezone minutes part was always 0 instead of the proper value.
Additionally fixed minor typos in comments.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Changes:
* Bash based runners/run-test-suite.sh runner replaced with a python runner
* Common util functions moved to util.py
* Fixed EOL issues in tests directory
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
This PR is the preparation of the #2946 issue.
(Date.parse() should handle UTC date string format)
Changes:
- Handle min/max values in ecma_date_parse_date_chars function instead of code duplication
- Parse special characters in the new ecma_date_parse_special_char function
- Additionally fixed two parsing issue: invalid formats should be refused
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
We should emit symbol for ecma_make_nan_value (void), because
it is used by unit-core/test-to-integer.c and test-to-length.c.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
ES5.1 15.9.1.15 Note1 defines that 24:00 is same as 0:00
of the next day. The spec explicitly doesn't mention that
24:01 should be invalid, but it should be self-evident.
(FireFox and Chrome also refuses times bigger than 24:00)
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
This change makes ecma_number_make_nan and ecma_number_make_infinity
always return constant value without any function call. Previously we
relied on compiler optimizations.
The ecma_number_t_accessor union is introduced to be able to access
float values as float and uint32_t (and doubles as double and uint64_t)
properly, without violating strict aliasing rules. There were many
copies of it, all of them were replaced to this new union.
Additionally ecma_number_make_nan should return QNaN instead of SNaN,
same value as C99 nan(""). Unfortunately calling nan("") here isn't
always optimal, because compilers sometimes generate constant returns,
sometimes function calls.
Before this change ecma_number_make_nan returned SNaN:
- double: 0x7FF0 0000 0000 0001 (sign:0, exponent: all 1 bits, fraction: 0...01)
- float: 0x7F8 00001 (sign:0, exponent: all 1 bits, fraction: 0...01)
After this change ecma_number_make_nan returns QNaN:
- double: 0x7FF8 0000 0000 0000 (sign:0, exponent: all 1 bits, fraction: 10..0)
- float: 0x7FC0 0000 (sign:0, exponent: all 1 bits, fraction: 10...0)
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
The implementation was incorrect for negative years and years bigger than 9999.
-1 was 000/ because the negative (year%10) was added to '0' character, years
bigger than 9999 was truncated to 4 digits.
ES5.1 15.9.1.15.1 defines extended years format with 6 digits, but toString()
and toISOString() sections don't mention anything about extended years. ES6
20.3.4.3 already clarifies that Date.prototype.toISOString() should use this
extended year format if it is necessary.
Changes:
- Date.prototype.toString() uses 4 digits for years by default, 5 or 6 if it
is necessary and put '-' sign for negative years, no sign for positive years.
Date.prototype.toString() was implementation dependent until ES9, but ES9
already specify exactly this format.
- Date.prototype.toISOString() uses fixed 4 digits for years 0 - 9999,
otherwise sign + 6 digits (extended years).
- Tests added for corner cases.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
ES5.1 15.9.4.2 specifies that Date.parse (string) has to accept
at least Date Time String Format: YYYY-MM-DDTHH:mm:ss.sssZ. But
the spec allows implementation-specific fallbacks. Additionally
ES5.1 15.9.1.15.1 specifies Extended years format, but isn't
explicitly required to be accepted by Date.parse. But ES6 already
clarified that Date.parse has to accept extended years format too.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
We should emit function symbols for these jerry-core functions:
- ecma_compare_ecma_strings is used by test-stringbuilder.c
- ecma_is_value_number is used by test-literal-storage.c
- ecma_date_time_within_day is used by test-date-helpers.c
- jmem_heap_alloc_block is used by test-jmem.c
- jmem_heap_free_block is used by test-jmem.c
- jmem_pools_alloc is used by test-poolman.c
- jmem_pools_free is used by test-poolman.c
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Changes:
- Implemented jerry_port_get_local_time_zone_adjustment on Windows
- Implemented jerry_port_get_current_time on Windows
- Run test262 tests on Windows in PST/PDT timezone
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
subprocess.Popen needs universal_newlines=True parameter
to open the file in text mode instead of binary mode.
With this fix readline() returns str instead of bytes.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Fixes#3142.
Changes:
- Disable LTO on Cygwin build because of a Cygwin-ld bug.
- Don't add "-z noexecstack" to ld because it is ELF specific option.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Disable 02.API-REFERENCE-create-context.c, because
it is based on pthread, and there is no pthread on Windows.
Disable 11.EXT-REFERENCE-AUTORELEASE.c and test-ext-autorelease.c
if compiler is MSVC, because MSVC doesn't support cleanup attribute.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
We should emit function symbols for ecma_get_magic_string
function, because it is used by unit-test-stringbuilder.c.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
If system allocator is used jerry_heap_stats_t.size
is always 0 and we can't get detailed heap statistics.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
MSVC doesn't support library constructor/destructor, let's disable
the related unittest and clarify the documentation and error messages.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
If manager_p->bytes_needed == 0, jerry_get_context_data() should return
NULL pointer. Additionally init_cb, deinit_cb and finalize_cb should be
called with NULL pointer in this case.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
We have to use jerry_get_string_size() to get the necessary buffer
size instead of jerry_get_string_length() and jerry_json_stringify().
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Changes:
* Bash based unittest runner replaced with a python runner
* Typo fixed in doctest cmake build system (python executable)
* run-tests.py prints error message if build fails
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Changes:
- gen-doctest.py: Use slashes in paths to make cmake happy.
- unit-doc/CMakeLists.txt: Don't add invalid arguments to MSVC and call gen-doctest.py properly.
- build.py: Build and install with cmake calls on Windows.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
- Added pylint and python-serial to apt-get-install-deps.sh.
- Make Jerry debugger import python serial module only if needed.
- Check-pylint.sh warns if pylint isn't installed.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
We should emit function symbols for ecma_date_day, ecma_is_value_string and
ecma_number_is_nan inline functions, because they are used by unit tests:
unit-core/test-date-helpers.c, test-literal-storage.c and test-string-to-number.c.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
MSVC doesn't support C99 VLA (variable-length array).
Use fixed size arrays in API examples and use JERRY_VLA macro
in unit tests to make these tests buildable on Windows too.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
MSVC doen't support __attribute__((unused)),
we should use JERRY_UNUSED macro instead.
Additionally removed the internal jrt.h include from
tests/unit-core/test-common.h which was layering violation.
It made JERRY_ASSERT unavailable, we should use TEST_ASSERT.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
The Math.pow implementation calls libm's pow. The ISO C and ES5.1
standards differ on some special cases of pow. jerry-libm is already
ES5.1 conform, but system libm libraries on Linux and Windows aren't.
Math.pow(NaN, +/-0.0) is NaN on Windows and Linux with system libm,
but should be 1.0 according to ES5.1 / 15.8.2.13.
This patch handles this special case in Math.pow instead of calling pow of libm.
JerryScript-DCO-1.0-Signed-off-by: Csaba Osztrogonác [email protected]
Csaba Osztrogonác