Initial support for mbed OS 5.1 #1318

matthewelse · 2016-09-02T10:40:24Z

Added target for the recently released mbed OS 5.1. So far we've tested JerryScript on the boards listed in targets/mbedos5/README.md (though it should work on the various other boards supported by mbed OS 5.1), and instructions are given in that README about how to build it.

cc: @thegecko @stephenkyle-ARM @janjongboom

zherczeg · 2016-09-05T06:51:37Z

Wow this is a huge amount of work! Am I see right that the code is in C++ and follows the style of mbed? What is the purpose of .mbedignore?

LaszloLango · 2016-09-05T07:04:59Z

Is it possible to generate the pin_defs directory? Not sure they should be in jerryscript repository.

matthewelse · 2016-09-05T09:19:58Z

@zherczeg the purpose of .mbedignore is to allow us to build the jerryscript sources using mbed OS 5's build tool 'mbed cli', without building things like jerry-main and jerry-libc.

@LaszloLango yes, it is possible to generate pin_defs (it follows the same directory structure as the mbed HAL) and there's a script that @stephenkyle-ARM wrote to generate it. I imagine that we could integrate it into the Makefile, though it would probably slow down the initial build quite considerably.

zherczeg · 2016-09-06T06:38:00Z

The truth is I am not in a favour of adding 100+ generated C files to the project. It would be great to figure something out for getting them (generating? downloading as sub-repository?)

janjongboom · 2016-09-06T08:14:02Z

@matthewelse We don't need to generate every pindef file when we move it to the Makefile right, can only generate it when building for a new target.

stephenkyle-ARM · 2016-09-06T09:52:23Z

Those pin_defs files were generated from the mbed HAL PinNames descriptions: https://github.com/ARMmbed/mbed-os/tree/master/hal/targets/hal

As @matthewelse says, I have a script that generates the pin_defs files, however it depends on using libclang and its python bindings for parsing C++. I haven't checked all the PinNames.h files to see if they could easily be parsed using something more hand-written, but that would probably be the way to go if we can't just include all the generated pins.js files in this project.

janjongboom · 2016-09-06T10:42:08Z

@stephenkyle-ARM Can we put them in a separate repo and then pull it in same way we pull in mbed-os via make getlibs? It'll be a PITA when we need to open a new PR against jerryscript every time a new target is added in mbed-os repo.

matthewelse · 2016-09-08T10:00:33Z

I've got a small update coming soon, which should fix a few issues I was having where jerry_parse would crash with non-trivial examples due to a lack of stack size. Howeber, I need to wait for ARMmbed/mbed-os#2646 to be merged first.

As for the target-specific (auto-generated) js code, I agree that it would be nice to be able to avoid having to make a new pull request every time we want to add support for a new target, If there's any way we can reduce the complexity of dependencies required (i.e. preferably no libclang if possible), it'd be nice to generate pins.js on the fly as an additional build step.

matthewelse · 2016-09-13T07:59:11Z

We now have legal approval to open source the port, so the DCO is now signed, and squashed.

CC: @janjongboom @thegecko @stephenkyle-ARM

LaszloLango · 2016-09-13T14:00:06Z

targets/mbedos5/mbed_app.json

+            "target.uart_hwfc": 0
+        }
+    }
+}


Please add a new line at the end of the file

Perhaps we shouldn't include an mbed_app.json file anyway, @matthewelse ?

Only reason I included it was to make sure that standard I/O on the NRF52 doesn't cause the program to hang if you aren't connected to the serial port (seems like hanging is the default behaviour in mbed OS for some reason)

akosthekiss

I've spotted some licensing issues, mentioned above.

In addition to that, I second @zherczeg 's opinion that houndred+ generated files should not be included in the repository. Is there any solution upcoming for that?

akosthekiss · 2016-09-15T06:00:15Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-launcher/source/launcher.cpp

@@ -0,0 +1,72 @@
+/* Copyright (c) 2016 ARM Limited. All rights reserved. */


Missing license

Fixed in d125e2c

akosthekiss · 2016-09-15T06:00:15Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-util/js_source.h

@@ -0,0 +1,11 @@
+/* Copyright (c) 2016 ARM Limited. All rights reserved. */


Missing license

Fixed in d125e2c

akosthekiss · 2016-09-15T06:00:15Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-util/logging.h

@@ -0,0 +1,13 @@
+/* Copyright (c) 2016 ARM Limited. All rights reserved. */


Missing license

Fixed in d125e2c

akosthekiss · 2016-09-15T06:00:15Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-util/wrappers.h

@@ -0,0 +1,84 @@
+/* Copyright (c) 2016 ARM Limited. All rights reserved. */


Missing license

Fixed in d125e2c

akosthekiss · 2016-09-15T06:00:15Z

targets/mbedos5/tools/js2c.py

+#!/usr/bin/env python
+
+# Copyright 2015-2016 Samsung Electronics Co., Ltd.
+# Copyright (c) 2016 ARM Limited. All rights reserved.


Fixed in d125e2c

akosthekiss · 2016-09-15T06:00:15Z

targets/mbedos5/tools/check_pins.sh

@@ -0,0 +1,31 @@
+#!/bin/bash
+
+# Copyright (c) 2016 ARM Limited. All rights reserved.


Fixed in d125e2c

akosthekiss · 2016-09-15T06:00:15Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-library-registry/source/wrap_tools.cpp

@@ -0,0 +1,94 @@
+/* Copyright 2014-2015 Samsung Electronics Co., Ltd.
+ * Copyright (c) 2016 ARM Limited. All rights reserved.


Fixed in d125e2c

akosthekiss · 2016-09-15T07:46:34Z

A comment/idea about the pindefs: As it is heavily dependent on the mbed-os module/library, it might not be a good idea to pre-generate and store them (like, anywhere; neither here, nor at an external place). If and when mbed-os gets any maintenance changes or extensions, it will be error-prone to keep them in sync. On the other hand, I understand that running the previously mentioned script that generates the js/pin_defs directory may take a long time to complete. However (and here comes the core message/idea): why do we have to generate all pins.js files when we build for a single given target? Cannot we have a modified version of that script that generates one pins.js and that's it? (Note: I also understand that it has dependencies that jerryscript itself has not (libclang and python bindings). However, the whole target has additional dependencies, even in terms of tools, e.g., the mbed cli, so this should not be a blocking point.)

An additonal note: I have the impression that the js2c.py file in this target does a bit more than its name suggests. It mixes the selection of the correct pins.js and creates the .c file with the array from all .js files (pins.js included). If there would be something like a gen_pins_js.py (see above), then js2c.py could be kept "clean" and kept in sync with js2c.py used by other targets (to avoid copy-pasting, cloning, and promote code reuse).

galpeter · 2016-09-15T08:31:18Z

targets/mbedos5/tools/js2c.py

+        FNULL = open(os.devnull, 'w')
+
+        try:
+            cmd = "dir /s /b /AD \"{}\"".format("TARGET_" + args.mcu)


I'm not sure that it's really a good idea to have a separate windows/linux cmd for searching for file. We can do this directly in python no need to call out to the system.

galpeter · 2016-09-15T08:31:18Z

targets/mbedos5/tools/js2c.py

+                print('\n'.join(js_target_list))
+                sys.exit(1)
+            else:
+                os.system("cp {} ./js/pins.js".format(js_target_list[0]))


Python has good support for file operations, please check the shutil module.

galpeter · 2016-09-15T08:31:18Z

targets/mbedos5/tools/js2c.py

+
+        try:
+            cmd = "dir /s /b /AD \"{}\"".format("TARGET_" + args.mcu)
+            intended_target = subprocess.check_output(cmd, shell = True, cwd=SRC_PATH, stderr=FNULL).strip()


Calling the subprocess methods with shell=True is not advised.

galpeter · 2016-09-15T08:31:18Z

targets/mbedos5/tools/js2c.py

+        print("JSHint picked up an undefined variable in your code.")
+        print("Are you using a PinName that doesn't exist for your platform?")
+        raise
+        sys.exit(1)


I don't think that the sys.exit will be ever called if we re-raise the exception before.

galpeter · 2016-09-15T08:31:18Z

targets/mbedos5/tools/js2c.py

+def exportOneFile(path, name):
+    # Prepend pins.js onto main.js, don't include pins.js itself.
+    code = ""
+    if name == "pins":


I would not put the pin name/value extraction here as it clearly not the same as the methods original purpose. Simply add a check where the method is called and do the extraction there (preferably as a method).

galpeter · 2016-09-15T08:31:18Z

targets/mbedos5/tools/js2c.py

+exportOneName('main')
+filenames = sorted(map(extractName, files), key=str.lower)
+for name in filenames:
+    if name != 'main' and name != 'pins':


Better way: if name not in ['main', 'pins']:

galpeter · 2016-09-15T08:31:18Z

targets/mbedos5/tools/js2c.py

+writeLine(fout, 'const char *jsmbed_js_magic_strings[] = {')
+comma = ","
+
+for (idx, (pinname, value)) in enumerate(pins):


I'm not a big fan of iterating over the same thing more than one times. We could extract the required data in one go. Then write it out when we need it. Also adding the comma manually (basically joining strings) makes me wondering if I can do it better.

In short, maybe something like this can be useful:

string_entries = [] length_entries = [] value_entries = [] for (idx, (name, value)) in enaumerate(pins): string_entries.append('"{}"'.format(name)) length_entries.append('{}'.format(len(name))) value_entries.append('{}'.format(value)) writeLine(fout, ',\n '.join(string_entries), 1) ... writeLine(fout, ',\n '.join(length_entries), 1) ... writeLine(fout, ',\n '.join(value_entries), 1)

galpeter · 2016-09-15T08:31:18Z

targets/mbedos5/tools/js2c.py

@@ -0,0 +1,276 @@
+#!/usr/bin/env python


I'm have the same view as @akosthekiss, the pin js file generation and extraction should be done in a different script and not in the js2c.py.

matthewelse · 2016-09-15T08:34:34Z

@galpeter I'm currently working on a new python script to generate pins.js on the fly, so that we can make use of the common js2c.py tool. As such, it's worth waiting for the time being before we do a thorough code review.

akosthekiss · 2016-09-15T09:21:10Z

@matthewelse, just a quick note: if it turns out that the common js2c.py does not quite suite your needs, or could be improved (not in a way that would mix in unrelated functionality but so that it could perform its task better, in a more generic or adaptable way, with less hard-coded stuff, etc), do propose changes to it. That might be useful for other targets as well.

akosthekiss · 2016-09-15T22:00:13Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-drivers/lib_drivers.h

+#include "jerryscript-mbed-drivers/InterruptIn-js.h"
+#include "jerryscript-mbed-drivers/DigitalOut-js.h"
+#include "jerryscript-mbed-drivers/setInterval-js.h"
+ #include "jerryscript-mbed-drivers/setTimeout-js.h"


Any reason for this extra space?

akosthekiss · 2016-09-15T22:00:13Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-drivers/source/DigitalOut.cpp

+
+
+DECLARE_CLASS_FUNCTION(DigitalOut, write) {
+    // check and unbox arguments


Most of the target code seems to use 2-space indentation, but this file uses 4 spaces. It would be nice to use a consistent style throughout the target.

Another style inconsistency is the placement of braces (together with declaration or on separate line).

akosthekiss

First round of review on the updated PR. Some comments related to style, some related to the use of jerry API, some are more general.

And one more (couldn't put a line comment in an empty file): what is tools/cmsis.h for? It's empty, and couldn't see it used.

akosthekiss · 2016-09-15T23:03:42Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-drivers/source/InterruptIn.cpp

+  ATTACH_CLASS_FUNCTION(js_object, InterruptIn, disable_irq);
+
+  return js_object;
+}


avoid files without NL at EOF

akosthekiss · 2016-09-15T23:03:42Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-drivers/source/DigitalOut.cpp

+}
+
+
+


style: too many empty lines

akosthekiss · 2016-09-15T23:03:42Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-drivers/source/DigitalOut.cpp

+    }
+    else {
+        LOG_PRINT_ALWAYS("invalid arguments in method write");
+        return jerry_create_undefined();


It might be better perhaps to return an error object with an error message (e.g., with jerry_create_error(JERRY_ERROR_TYPE, "...")), instead of logging and returning undefined.

akosthekiss · 2016-09-15T23:03:42Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-drivers/source/assert.cpp

+  {
+    LOG_PRINT_ALWAYS("ERROR: Script assertion failed\n");
+    exit(1);
+    return jerry_create_undefined();


exit and then return? Looking at the code, the compiler should not complain even if return would be omitted.

akosthekiss · 2016-09-15T23:03:42Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-event-loop/BoundCallback.h

+
+    BoundCallback(Callback<R(A0)> cb, A0 a0) : a0(a0), cb(cb) { }
+
+    void call() {


another file with style inconsistencies

akosthekiss · 2016-09-15T23:03:42Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-event-loop/EventLoop.h

+    static EventLoop instance;
+
+ public:
+    static EventLoop& getInstance() {


ditto (style: indent, braces)

akosthekiss · 2016-09-15T23:03:42Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-library-registry/registry.h

+    static LibraryRegistry instance;
+
+public:
+    static LibraryRegistry& getInstance() {


ditto (style)

akosthekiss · 2016-09-15T23:03:42Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-event-loop/EventLoop.h

+    Callback<void()> wrapFunction(jerry_value_t f) {
+        if (!jerry_value_is_function(f)) {
+            LOG_PRINT_ALWAYS("invalid function passed\r\n");
+            exit(1);


Is this really what we want? That exit signals some bad smell to me. Shouldn't these four lines just be replaced with an assert(jerry_value_is_function(f));, or something similar?

akosthekiss · 2016-09-15T23:03:42Z

targets/mbedos5/tools/generate_pins.py

@@ -0,0 +1,214 @@
+"""


Missing Apache 2.0 license header

matthewelse · 2016-09-16T12:05:00Z

@galpeter and @akosthekiss - I think the most recent commits should address the suggestions you had.

akosthekiss

Second round of comments added.
I still miss the explanation for the empty tools/cmsis.h f file.

akosthekiss · 2016-09-16T14:00:31Z

targets/mbedos5/Makefile

+MBED_CLI_FLAGS += -t GCC_ARM
+
+.PHONY: all js2c getlibs rebuild library pins
+all: source/jerry_target.h source/pins.cpp .mbed ../../.mbedignore


source/jerry_target.h is bad reference. source/jerry_targetjs.h might be better

akosthekiss · 2016-09-16T14:00:31Z

targets/mbedos5/Makefile

+	mbed target $(BOARD)
+	mbed compile $(MBED_CLI_FLAGS) --library
+
+rebuild: clean all


This is dangerous. AFAIK, make makes no guarantees about the order of the execution of the dependencies. Especially if you run it with -j, then all and clean will happen in parallel.

akosthekiss · 2016-09-16T14:00:31Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-drivers/source/I2C.cpp

+    }
+
+    LOG_PRINT_ALWAYS("ERROR: Unexpected number of arguments to I2C.read, expected 1, 3 or 4.");
+    return jerry_create_undefined();


a jerry_create_error would fit here as well

akosthekiss · 2016-09-16T14:00:31Z

targets/mbedos5/mbed-events.lib

@@ -0,0 +1 @@
+https://github.com/ARMmbed/mbed-events/#a4ba1b08ef90e2b3b6d442696f4d80a8587494e3


According to https://docs.mbed.com/docs/mbed-os-handbook/en/5.1/dev_tools/cli/, .lib files are not to be hand-managed: "Warning: mbed CLI stores information about libraries and dependencies in reference files that use the .lib extension (like lib_name.lib). While these files are human-readable, we strongly advise that you don’t edit these manually - let mbed CLI manage them instead." Do we need to check this file in the repository, or should they better be added to .gitignore?

These .lib files definitely need to be in source control, otherwise mbed cli doesn't know what libraries to fetch.

akosthekiss · 2016-09-16T14:00:31Z

targets/mbedos5/mbed-os.lib

@@ -0,0 +1 @@
+https://github.com/ARMmbed/mbed-os/#bdab10dc0f90748b6989c8b577771bb403ca6bd8


same question for this file

akosthekiss · 2016-09-16T14:00:31Z

targets/mbedos5/tools/generate_pins.py

@@ -0,0 +1,226 @@
+"""
+getpins_js.py


Name is incorrect. Moreover, I suggest to let this file start just like other python files in the project:

#!/usr/bin/env python # Copyright 2016 Samsung Electronics Co., Ltd. # Copyright 2016 University of Szeged. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License.

That is,

no repetition of the file name, so you wont have to keep it in sync with actual file name,

let first line be a hashbang line,

put license in comment, not in docstring.

(of course, with a different copyright notice)

akosthekiss · 2016-09-16T14:00:31Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-util/wrappers.h

+#define CHECK_ARGUMENT_COUNT(CLASS, NAME, EXPR) \
+  if (!(EXPR)) { \
+    LOG_PRINT_ALWAYS("ERROR: wrong argument count for %s.%s, expected %s.\n", # CLASS, # NAME, # EXPR ); \
+    return false; \


This code is suspicious. This macro is used in functions, which return jerry_value_t. Returning false here (and printing) is most probably not the right way. (But returning with jerry_create_error might be the right thing.)

You're correct - this code is a hangover from an older version of the jerryscript API, and GCC is casting the false value to jerry_value_t, which is definitely bad. I'll fix this, and similar macros too.

akosthekiss · 2016-09-16T14:00:31Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-drivers/source/DigitalOut.cpp

+        return jerry_create_undefined();
+    } else {
+        const char* error_msg = "Invalid arguments in method write";
+        return jerry_create_error(JERRY_ERROR_TYPE, reinterpret_cast<const jerry_char_t*>(error_msg));


BTW, couldn't the CHECK_ARGUMENT_ macros be used in this file's functions as well? (Once their return value is fixed, of course.)

@akosthekiss you're right, the only reason it's like this is because DigitalOut.cpp was auto-generated using a script I wrote at some point, though for consistency since we're putting it into source control, I'll modify it to be consistent with the other driver sources.

akosthekiss · 2016-09-16T14:11:35Z

targets/mbedos5/.gitignore

+mbed-events
+.build
+.mbed
+mbed_settings.py


Cannot find this file, is it needed in .gitignore?

mbed_settings.py is created by mbed cli in mbedos5, but is specific to individual users' machines, so shouldn't be under version control.

matthewelse · 2016-09-16T14:13:32Z

@akosthekiss the empty cmisis.h file is to simplify the process of parsing PinNames.h files in the mbed source tree (prevents a huge number of header files being included)

akosthekiss · 2016-09-16T14:25:09Z

@matthewelse Then I'd suggest putting a proper content in the cmsis.h, with header + a comment with explanation like /* Intentionally empty replacement header for the purposes of the generate_pins.py tool */

akosthekiss · 2016-09-18T12:06:12Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-drivers/source/DigitalOut.cpp

+/**
+ * DigitalOut#is_connected (native JavaScript method)
+ *
+ * @returns 0 if the DigitalOut is set to NC, or 1 if it is connected to an


Is the 0/1 return value intentional? JS does have a boolean type. Why not return true/false?
(Not insisting though, just asking.)

There's no particular reason not to return true/false, this is simply matching what is returned by the mbed API.

OK with me to keep it integer then

akosthekiss · 2016-09-18T12:37:55Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-drivers/source/I2C.cpp

+    }
+
+    const char* error_msg = "ERROR: Unexpected number of arguments to I2C.read, expected 1, 3 or 4.";
+    return jerry_create_error(JERRY_ERROR_TYPE, reinterpret_cast<const jerry_char_t*>(error_msg));


Couldn't this be replaced with CHECK_ARGUMENT_COUNT(I2C, read, (args_count == 1 || args_count == 3 || args_count == 4));, moved to the start of the function?

akosthekiss · 2016-09-18T12:37:55Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-drivers/source/I2C.cpp

+ */
+DECLARE_CLASS_FUNCTION(I2C, frequency) {
+    CHECK_ARGUMENT_COUNT(Ticker, frequency, (args_count == 1));
+    CHECK_ARGUMENT_TYPE_ALWAYS(Ticker, frequency, 0, number);


Ticker is not right here I guess, should be I2C.

akosthekiss · 2016-09-18T12:37:55Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-drivers/source/I2C.cpp

+    }
+
+    const char* error_msg = "ERROR: Unexpected number of arguments to I2C.write, expected 1, 3 or 4.";
+    return jerry_create_error(JERRY_ERROR_TYPE, reinterpret_cast<const jerry_char_t*>(error_msg));


(same macro approach here as for I2C.read)

akosthekiss · 2016-09-18T12:37:55Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-library-registry/source/wrap_tools.cpp

+jsmbed_wrap_register_global_function(const char* name,
+                          jerry_external_handler_t handler)
+{
+  jerry_value_t global_object_val = jerry_get_global_object();


(style) This single file seems to have retained the 2-space indentation. Stick to one style.

akosthekiss · 2016-09-18T12:37:55Z

targets/mbedos5/tools/cmsis.h

@@ -0,0 +1,4 @@
+/*


This file only got a body, not a license header, as requested. Please add a proper Apache 2.0 license header.

akosthekiss · 2016-09-18T12:37:55Z

targets/mbedos5/tools/generate_pins.py

+#!/usr/bin/env python
+
+# Copyright (c) 2016 ARM Limited
+


This is not exactly the header found in other files and suggested in previous review. The missing # seems like a minor technicality, but most probably we will have an automated license checker set up soon, and it will choke on such differences.

akosthekiss · 2016-09-18T12:37:55Z

targets/mbedos5/Makefile

+clean:
+	rm -rf .build/$(BOARD)
+
+js2c: js/main.js js/flash_leds.js pins


Why is pins a dependency if pins.js is ignored? (BTW, why is pins.js ignored, will it be not needed to declare the LED1..4 variables?)

pins.js is ignored, since it is incorporated into the binary as an array of magic strings, to reduce memory consumption.

Q1: Is jshint the only reason for pins.js then?
Q2: Isn't pins a superfluous dependency for js2c then?

Q1: yes
Q2: yes (fixed in soon to be pushed commit)

akosthekiss

I think we are getting close. I'm running out of comments. These are hopefully my final comments before approval.

akosthekiss · 2016-09-20T16:46:15Z

targets/tools/js2c.py

@@ -21,8 +21,12 @@
 import os
 import re

+import argparse
+
+special_chars = re.compile('[-|\\\\|\\?|\\\'|".]')


This regex does not seem right. Within character classes, you don't need alternation (actually, special characters lose their meaning). And using raw strings could help to get rid of lots of backspace escaping. A quick try in python shell gave:

>>> special_chars = re.compile(r'[-\\?\'".]') >>> special_chars.sub('_', "x-x?x'x\"x.x\\x") 'x_x_x_x_x_x_x'

(If you want to replace dash, backslash, question mark, single and double quotes, and period.)

akosthekiss · 2016-09-20T16:46:15Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-library-registry/source/wrap_tools.cpp

+
+    if (jerry_value_has_error_flag(set_result)) {
+        is_ok = false;
+        LOG_PRINT_ALWAYS("Error: register_native_function failed: [%s]\r\n", name);


Not sure what"register_native_function" is referring to. This function is named "jsmbed_wrap_register_global_function", perhaps that was intended.

akosthekiss · 2016-09-20T16:46:15Z

targets/mbedos5/jerryscript-mbed/jerryscript-mbed-library-registry/source/wrap_tools.cpp

+    if (!(jerry_value_is_function(reg_function)
+        && jerry_value_is_constructor(reg_function))) {
+        is_ok = false;
+        LOG_PRINT_ALWAYS("Error: create_external_function failed !!!\r\n");


Also a bit misleading error message missing the "jerry_" prefix.

(BTW: Any intention to get error messages even more consistent? Current strings use "ERROR" vs "Error" and "!!!" vs "!" vs "." vs nothing and "\r\n" vs "\n\r" vs "\n".)

akosthekiss · 2016-09-20T16:51:33Z

(One technical note: right now, there are 30 patches on top of each other. Before landing can happen, it would be nice to have all changes squashed, amended with a consolidated commit message, and rebased on latest master. Actually, we have everything in place to do that by the final reviewer/maintainer, but with 30 patches, it'd be better if you could do that, not to get anything wrong.)

matthewelse · 2016-09-21T09:23:06Z

@akosthekiss AFAICT this should resolve the last few issues you had. Thanks for the code review - the code is in much better shape thanks to your help 👍.

I've also gone ahead and squashed the commits together.

LaszloLango added development jerry-port labels Sep 2, 2016

matthewelse force-pushed the mbed-os-5 branch 2 times, most recently from e9ec9a0 to cf3d478 Compare Sep 5, 2016

matthewelse mentioned this pull request Sep 8, 2016

Add RTOS configuration for default stack size ARMmbed/mbed-os#2646

Closed

matthewelse force-pushed the mbed-os-5 branch from c0990ef to 2cd50ba Compare Sep 13, 2016

matthewelse force-pushed the mbed-os-5 branch 4 times, most recently from 19c9daf to 5965621 Compare Sep 13, 2016

LaszloLango reviewed Sep 13, 2016
View changes

matthewelse force-pushed the mbed-os-5 branch 4 times, most recently from fc80cdd to 1589fd7 Compare Sep 14, 2016

akosthekiss requested changes Sep 15, 2016

View changes

galpeter requested changes Sep 15, 2016

View changes

matthewelse force-pushed the mbed-os-5 branch from d125e2c to 1f3f152 Compare Sep 15, 2016

matthewelse force-pushed the mbed-os-5 branch from 28b02ea to 79f2ba1 Compare Sep 15, 2016

akosthekiss reviewed Sep 15, 2016

View changes

akosthekiss reviewed Sep 16, 2016

View changes

akosthekiss reviewed Sep 18, 2016

View changes

matthewelse force-pushed the mbed-os-5 branch from ce60eed to 4fa929f Compare Sep 19, 2016

akosthekiss reviewed Sep 20, 2016

View changes

matthewelse force-pushed the mbed-os-5 branch from 292b491 to 16c8de3 Compare Sep 21, 2016

matthewelse force-pushed the mbed-os-5 branch from 16c8de3 to

		@@ -0,0 +1,72 @@
		/* Copyright (c) 2016 ARM Limited. All rights reserved. */

		@@ -0,0 +1,11 @@
		/* Copyright (c) 2016 ARM Limited. All rights reserved. */

		@@ -0,0 +1,13 @@
		/* Copyright (c) 2016 ARM Limited. All rights reserved. */

		@@ -0,0 +1,84 @@
		/* Copyright (c) 2016 ARM Limited. All rights reserved. */

		@@ -0,0 +1,31 @@
		#!/bin/bash

		# Copyright (c) 2016 ARM Limited. All rights reserved.

		@@ -0,0 +1,94 @@
		/* Copyright 2014-2015 Samsung Electronics Co., Ltd.
		* Copyright (c) 2016 ARM Limited. All rights reserved.



		DECLARE_CLASS_FUNCTION(DigitalOut, write) {
		// check and unbox arguments


		BoundCallback(Callback<R(A0)> cb, A0 a0) : a0(a0), cb(cb) { }

		void call() {

		@@ -0,0 +1 @@
		https://github.com/ARMmbed/mbed-events/#a4ba1b08ef90e2b3b6d442696f4d80a8587494e3

		@@ -0,0 +1 @@
		https://github.com/ARMmbed/mbed-os/#bdab10dc0f90748b6989c8b577771bb403ca6bd8

jerryscript-project / jerryscript Public

Initial support for mbed OS 5.1 #1318

Initial support for mbed OS 5.1 #1318

Conversation

matthewelse commented Sep 2, 2016

zherczeg commented Sep 5, 2016

LaszloLango commented Sep 5, 2016

matthewelse commented Sep 5, 2016

zherczeg commented Sep 6, 2016

janjongboom commented Sep 6, 2016

stephenkyle-ARM commented Sep 6, 2016 • edited

janjongboom commented Sep 6, 2016 • edited

matthewelse commented Sep 8, 2016 • edited

matthewelse commented Sep 13, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

akosthekiss left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

akosthekiss commented Sep 15, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

matthewelse commented Sep 15, 2016

akosthekiss commented Sep 15, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

akosthekiss left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

matthewelse commented Sep 16, 2016

akosthekiss left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

matthewelse commented Sep 16, 2016

akosthekiss commented Sep 16, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

stephenkyle-ARM commented Sep 6, 2016 •

edited

janjongboom commented Sep 6, 2016 •

edited

matthewelse commented Sep 8, 2016 •

edited