fanoush

Thanks, I guess one just needs to not forget clicking "Flex" checkbox on ordering page (?)

Just to let you know that I managed to attach to both cpus via openocd and stlink v2 clone over SWD and managed to backup flash from icdi cpu and...

And BTW this may be useful too if you have two launchpads http://processors.wiki.ti.com/index.php/Stellaris_LM4F120_LaunchPad_Debug_How_To then one can debug one launchpad with another when the ICDI chip is not working. However jtag...

BTW, was checking how to avoid the export table altogether so one could possibly link any code from existing espruino binary and found this https://stackoverflow.com/questions/35183874/building-a-two-part-firmware-image-using-gcc-toolchain Maybe it is overkill but...

Well, since nrf52 has native floats I tried to enable it in the compiler and it seems to work. This example compiles fine and the output is same with calling...

However looks like I cannot put such single precision floats anywhere into signature like e.g. // int mandel(float, float) it compiles fine but breaks later when the method is called...

Looks like there are several subtopics: **double in compiledC support** (+) datatype already defined, is native JS type so good for passing without conversion (-) slow because of soft floats....

I checked Float32Array and indeed it uses 32bit float type. The only places where 'float' type is used is when [loading](https://github.com/espruino/Espruino/blob/master/src/jsvariterator.c#L378) and [storing](https://github.com/espruino/Espruino/blob/master/src/jsvariterator.c#L442) the value. And everywhere else only JsVarFloat...

> I guess a simple addition would be to export the float2double, double2float functions from Espruino such that you could at least write code using floats and then get/return the...

Oh, and BTW I found the bug why the power drain fix was not effective - pointer arithmetics :-) `return ((NVIC_Type *)0xE000E000UL+0x0100UL);` `return ((NVIC_Type *)(0xE000E000UL+0x0100UL));` the first line is wrong...