ClientContext::_wait_for_sent() could dereference a TCP's _pcb after
the connection was dropped by the OS, resulting in a crash.
Move the connection dropped check to catch this case, and replace
a fixed millisecond delay() with a yield and timeout value to minimize
wasted time when transmission completes.
* - Add assert in HelloMesh.ino for invalid transmission status.
- Make uint64ToString and stringToUint64 methods into stand-alone type conversion functions.
- Add getters and setters for requestHandler and responseHandler.
- Polish HelloMesh.ino code by adding networkIndex as networkFilter loop variable and switching networkFilter definition position.
- Add initial WiFi.disconnect() in HelloMesh.ino setup() function to ensure smooth WiFi operation.
- Add latestTransmissionSuccessful() convenience method.
- Change default WiFi mode to WIFI_STA and improve handling of WiFi mode (fixes issue #5071).
- Add checks to methods that change AP properties to avoid unnecessary AP restarts.
- Add getter for ESP8266WiFiMesh SSID and getters and setters for ESP8266WiFiMesh settings related to hidden SSID usage, max station connections allowed per AP and WiFi timeouts.
- Make waitForClientTransmission method use more accurate timekeeping.
- Improve type usage.
- Improve comments.
- Update README.md, keywords.txt and library.properties.
* Make getter and setter order consistent throughout code.
* - Fix active AP getting turned off when calling begin().
- Fix crash bug due to WiFiServer duplication when using the ESP8266WiFiMesh copy constructor with the AP controller as argument, under certain circumstances.
* - Move non performance-sensitive Strings to flash memory to save RAM.
- Add comments explaining F(), FPSTR() and PROGMEM.
- Fix README.md formatting.
* Remove the uint64ToString and stringToUint64 methods from the ESP8266WiFiMesh class since they are now stand-alone functions in the TypeConversionFunctions files.
* Change the minimum valid argument value of the setMaxAPStations method to 0, since this value is also supported by the ESP8266.
* Fix compiler warning.
* Move ICACHE_* to unique sections, local LD script
Similar to PROGMEM changes, make the name of each ICACHE_* variable/fuction
unique to avoid issues with section conflicts.
Also rename the generated LD linker script to avoid issue with older copies
of the eagle.app.v6.common.ld which were generated by the build process
in a global directory before being moved to the {build.path}. The linker
would use the older, generated *.ld file instead of the generated one, which
would lead to runtime failures on some systems and cause the VTABLE location
to not correspond to the IDE menu selection.
Fixes#5115, and is an update to #5117 and #5116.
* Update boards.txt.py and platform.io build
* Add full debug builds for all tests
In order to ensure the -DEBUG builds don't fail with checked-in code, add
two jobs which build even/odd with full debug flags to Travis.
Fixes#5143
* Add in dump of platform.txt to verify build options
* Fix BASH comparison of strings
* Fix debug-mode compile error in ESP8266AVRISP
* Add -Wextra to CI (matches IDE "All" warnings)
* Allow cipher specification for BearSSL
BearSSL has many more ciphers than axTLS, but they are more compute intensive
and slower. Add an option to use only the same, limited security, axTLS ciphers
as well as allow users to specify any suite of ciphers they want using standard
BearSSL formats.
Fixes#5110
* Rename methods to avoid axtls references.
* Allow std::vector to set a list of allowed ciphers
For C++ afficionados, allow std::vectors to be passed in to the setCipher()
routine.
The BearSSL object will now keep a copy of any set ciphers and free on object
destruction. These custom lists should normally only be 1-4 entries long, so it
is not expected to be a memory hog having this extra copy.
* Allow GPIO 9 and 10 for waveform generation
While most ESP8266 modules use quad-io mode for their SPI flash ROM,
there are some which only use dual-io mode. Allow the unused pins
(GPIO 9 and 10) to have waveforms generated on them. Should the user try
this on a quad-io mode board, expect very bad things to happen.
* Add variant for 8285 to init GPIO 9/10
The 8285 only has 2-bit flash IO, so the other two pins can be used as
inputs (9/10). Set them to input to mirror the way other pins are set
up.
* Update waveform gen to only allow 9/10 on 8285
Update the common.h in both generic (remove TODO, it's done!) and the
8285 variant to make isFlashInterfacePin macro correct. Use that macro
to disable pins 9 and 10 in the common, non-8285 case.
As part of the "clear connection configuration for reused objects"
patch, a ::stop would reset the self-signed, trust anchors, etc.
WiFiClient, unfortunately, calls ::stop as part of the connection
process, so all of these settings were lost.
Now only clear the connection settings on ::stop if we've already
been connected.
Also update the github public key which changed yet again.
Fixes#5086
Sphinx 1.8.0 seems to always fail to build the docs due to interactions
with doctest.py. This means no CI builds can pass and no new PRs can be
applied.
Fix the version at 1.7.9 for now until the dependencies get updated to with 1.8.
The error seen is:
File "/home/travis/.local/lib/python2.7/site-packages/sphinx/highlighting.py", line 26, in <module>
from sphinx.ext import doctest
SyntaxError: unqualified exec is not allowed in function 'run' it contains a nested function with free variables (doctest.py, line 97)
The full traceback has been saved in /tmp/sphinx-err-e8JyUQ.log, if you want to report the issue to the developers.
Please also report this if it was a user error, so that a better error message can be provided next time.
A bug report can be filed in the tracker at <https://github.com/sphinx-doc/sphinx/issues>. Thanks!
With the changes in PR #5117, blink and other examples compile but the 8266
gets stuck in a reset loop. Undo the changes from that commit until we can
figure out the root cause and fix it.
boards generator updates:
* simplified mapping description: only flash and spiffs sizes are needed
* some ldscripts are renamed due to autogenerated names
* +2M/0, +2M/512K spiffs, +4M0
* reduce length of hidden strings in boards.txt (#5100, https://github.com/arduino/arduino-builder/issues/284)
* give more details in ldscripts (address, size, +rf-cal, +sdk-wifi-settings)
* Add information about the max_connection constructor argument.
* Correct information about max soft-AP connections.
* Add information about what happens when max_connection is reached.
* Correct terminology
* - Correct the valid range of max_connection to 1 - 8.
- Include reference.
* add stubs for more exception throw calls
Fixes https://github.com/esp8266/Arduino/issues/3358
* libc: make putc_r implementation weak
newlib provides its own implementation of _putc_r, which will call
_write_r (possibly after buffering). Make our implementation weak to
allow using the one from newlib.
Fixes https://github.com/esp8266/Arduino/issues/4630
* libc: fix incorrect return value of _write_r call
Should return number of bytes written, actually returned zero. This
resulted in std::cout going into failed state after the first write.
* tests: add test for output to std::cout
arduino-builder 1.3.25 (shipped with Arduino 1.8.5) forces full
recompilation when any file in the core directory is modified. Avoid
full recompilation by placing generated ld script into build
directory, not source directory.
Also fix an issue where git version description would not be generated
if there were spaces in build path.
In tools menu, boards belonging to each platform are grouped together.
The title of the groups is the board package name. When switching
between versions, mentioning the version number in package name helps
identify the installed version.
Rename board group from "ESP8266 Modules" to "ESP8266 Boards", to be
more similar to other platforms, and to reflect the fact that most
entries included are boards, not modules.
As an extra change, when building the boards manager package, set
version number in platform.txt based on git tag name. This saves one
step when creating the release — don't need to update platform.txt
twice.
Closes https://github.com/esp8266/Arduino/issues/5007.
The memory allocation failure message was accidentally stored in RAM
and not in PROGMEM.
panic() did not place the __FILE__ string in PROGMEM, either.
Move both to PROGMEM, save ~64 bytes of heap (depends on size of path
of the Arduino core library).
__FILE__ is used to name the segments used for each PROGMEM constant,
but __FILE__ may have a space in it. This would cause compilation
errors.
Add quotes around the entire segment name to work around this.
According to the GCC man page, __section__ attributes should only be used
for global variables. However, the PROGMEM and ICACHE_RODATA macros use
this variable decorator even for local variables. Most of the time it works,
but when a static or inlined function tries to use a PROGMEM/PSTR/etc.
variable the compiler can throw an error like:
error: XXX causes a section type conflict with YYY
Change the PROGMEM macro to emit a section name that is unique (a combo
of the file, line, and counter variables to ensure uniqueness). The
standard linker script will place them properly in .IROM without
any changes.
Fixes#5036 and others.
* Implementation of Functional and Scheduled option in Ticker lib
* Update example formatting
* More example updates
* More updates to example
* More updates to example
* Analog input voltage can be 3.3V
According to this tutorial and my personal test with my Amica, 1.0V delivers 320 and 3.3V delivers 1024 at 10-bit resolution. I guess 3.3V is possible.
* Update reference.rst
Added warnings.
* Update reference.rst
* Update reference.rst
requested changes
* ci: work around tagging issue in Travis CI deployment
When `draft: true` is set, incorrect commit/tag information is sent to
Github. Override tag/target fields for correct behavior.
Ref. https://github.com/travis-ci/travis-ci/issues/9852
* doc: clarify some points in the release process
