sa666666 / harmonycart Goto Github PK

Currently, a CRC error after downloading a chunk of data halts the download. With testing, CRC errors are not repeatable and typically, if the user tries the download several times, this results in a successful download. Occasionally a board needs hits CRC errors often and the user needs to do dozens of retries to get a successful download.

The software allows changing the retry count up to 3 I believe, but I think the default retry count should be much higher than it is, at least 10, and should allow the user to select higher values still. Hopefully this would result in a much higher probability of download success.

Occasionally, after a CRC error, the software will lock up. In some instances, the only way to get the software to respond again is to unplug the board and plug it back in.

I have a device that comes up in Windows as COM257, and the software would not detect it.

The problem seems to be that only com ports 1-256 are being searched. See this function:

const StringList& SerialPortWin32::getPortNames()
{
myPortNames.clear();

for(int i = 1; i <= 256; ++i)
{
TCHAR strPort[32] = {0};
sprintf(strPort, "COM%d", i);

if(openPort(strPort))
{
  uInt8 c;
  int n = receiveBlock(&c, 1);
  if(n >= 0)
    myPortNames.push_back(strPort);
}
closePort();

}

In practice I have seen com ports above 1000, even back to Windows 2000. Maybe upping the limit to 1024 would help.

The normal pull request isn't working here. The lpc21isp software used within Harmonycart has a newer version, 1.97, and relevant code changes should be brought into this project: https://sourceforge.net/projects/lpc21isp/files/lpc21isp/1.97/

Though not perfect, this version seems more reliable than older versions.

The MacOS version of the software does not work in Catalina, since Apple went 64-bit only.

Maybe a new version of the software should be 64-bit across the board?