If the first sending was not successful it is better to abort. Then we get a return value of -1 (instead of -2 if the 2nd call will also fail) and we do not need to block even longer, with another timeout.
Since the HID library adds some general HID definitions and structs, it is very useful to also include this in a custom library.
If you dupe the information you have problems of redeclaration.
Having those defintions inside the official HID library is essential for a good working additional HID API.
This commit just shows other that this option is available and not implemented.
You may use this to determine the Led Lights state of a keyboard or transmit data via RAWHID from the PC.
Quick usage guide:
int length = ((setup.wValueH << 8) | setup.wLength);
USB_RecvControl((uint8_t* data, USB_EP_SIZE); // Needs to be splitted into USB_EP_SIZE packets, not shown here. See HID Project.
Alternatively we can only pass the wIndex to getDescriptor but I suggest to just pass the pointer aka reference of the whole setup.
In guess (havent tested this) that this results in more or less the code size but its a) idential with the other functions and b) we late have more flexibility here.
The Code got a quick SerialKeyboard.ino test
The iterations in the for loop also use unsigned and the setup struct etc as well.
There was no change in HID required since we just init the inherited variables via constructor and the type is never mentioned.
The field is now built on-the-fly on the stack and sent over USB.
This change increase Flash usage and decrease SRAM usage:
before: 6114 / 241
after: 6152 / 216 (removed HIDDescriptor field)
delta: +38 / -25
SRAM is a much more scarse resource and this change free up to
about 10% of the "base" usage.
The check for available slot in PluggableUSB is done on the endpoint
and not on the number of plugged modules.
The modulesCount field is no longer useful and it has been removed.
Cristian Maglie
Federico Fissore
agdl
Sandeep Mistry
Nico
Martino Facchin