RE: [PATCH v2 3/4] rsp: add 'E' to escaped characters

[Ciaran Woodward <ciaranwoodward@xmos.com>]

Hi Tankut,

> Add 'E' to the list of escaped characters when sending/receiving
> binary data.  This is a preparation for the next patch, to be able to
> distinguish an error response from binary data that starts with 'E'.

I wonder if there is a better way to do this, given that this is for a totally
new packet. My concerns are twofold:

1. The other 'escaped' characters are part of the RSP packetization layer,
   which is conceptually below the message processing. This new packet is
   looking at both escaped and non-escaped message in order to figure out
   which type of response it is looking at, which is (in my opinion)
   unnecessary layer crossing.

2. The idea of allowing this list to grow when there are other possible
   solutions is just less scalable if other new binary packets are added
   in the future.

Instead, I would propose a similar approach to other packets which would
have the potential for 'ambiguous' replies, such as 'qCRC', 'qMemTags' etc.

That is: Always have a leading byte (not just in the error case) in order
to disambiguate what type of reply the message contains.

So (for example) the message is something more like:

'E NN' for an error

'd XX...' for binary data

Does that sound reasonable?
(Note that I am not a maintainer, so would appreciate other opinions!)

Thanks,
Ciaran