WriteRead using I2C and MAX7311

[dotnet]

WriteRead using I2C and MAX7311

[shaggygi]

I didn't look into this much, but you might need to set the buffer size to 1. Not completely sure. Also, make sure the BusId is correct. I believe it should be 1.

var device = new System.Device.I2c.Drivers.UnixI2cDevice(new I2cConnectionSettings(BusId, address));
device.Write(new byte[] { 0x00 });
byte[] readBuffer = new byte[1];
device.Read(buffer);

[joperezr]

Because I2C protocol only uses 1 data pin, there is not a method like the TransferFullDuplex that we have in SPI. Internally what we are doing is to just do ioctl on the I2c device and send the information, so I would expect that as long as you set the right size of the read and write buffers that it should work correctly. Did you try what @shaggygi suggested?

[shaggygi]

I tested this (only on Unix) and can repro this when sending Write or Read commands with an empty array or Span similar in @bigdnf example.

@joperezr I don't understand why the ioctl wouldn't take/ignore 0 buffer length, but I guess it is expecting something when doing a write/read function. To fix this, we could put a check in ReadWriteInterfaceTransfer and FileInterfaceTransfer to just return if writeBufferLength and readBufferLength are both 0.

I don't like this as it is a little more overhead for checking on every Write/Read call and if someone used an empty array/Span... no error would be shown, the buffer would still be empty and you'd probably see another GitHub issue created stating the I2C method wasn't working.

In addition, when cleaning/updating the I2C and SPI summaries in PRs #143/#144, I added comments for buffer size when reading. This would probably help a little.

Close issue as is?

[dominikjeske]

Thanks for info - I'm currently on business trip so I can't test anything. I will do it after return and close issue if I can fix it after your suggestions.

[joperezr]

@bigdnf have you had the chance if @shaggygi suggestions work in your scenario?

I tested this (only on Unix) and can repro this when sending Write or Read commands with an empty array or Span similar in @bigdnf example.

I don't see that in the example @bigdnf is using an empty array or Span, so that might be a different issue.

To fix this, we could put a check in ReadWriteInterfaceTransfer and FileInterfaceTransfer to just return if writeBufferLength and readBufferLength are both 0.

I believe that this is an actual bug if we aren't checking the size, and if so, I would probably even throw an exception maybe if the read buffer or writebuffer is 0 as this shouldn't be allowed.

[dominikjeske]

@joperezr sorry for my delay with testing this - I will try to give feedback as soon as possible.

[joperezr]

no worries at all, just wanted to check if that was something you still had in your plans 😄

[krwq]

I think the correct fix should be to throw a descriptive exception when someone passes 0-length buffer to Read or am I missing something?

[shaggygi]

@krwq The original issue was because there were no bytes allocated to perform read. Your suggestion with having a better exception description would help in this case. Other than that, I would label issue as stale and close.