Variable length decoding in place, where ends the data? about nanocobs HOT 4 CLOSED

Ah, the _inplace decode and encode are only guaranteed to work if the entire payload is <= COBS_INPLACE_SAFE_BUFFER_SIZE which you'll see is small. _inplace encoding is more interesting, because it's an optimization that guarantees you don't need an extra buffer. The big difference with the _inplace API is that it lets you encode in-place, which is possible only with very constrained payloads. cobs_decode_inplace is there for parity.

If you're only looking at decoding, and you have a 10,000 byte encoded buffer, you can still decode it "in-place" by simply calling cobs_decode with the out_dec and enc pointers pointing to the same address: byte 0 of the encoded buffer.

As I type this up I realize the naming conventions aren't great, sorry about that. I'll brainstorm better names, also if you have any suggestions I'm all ears :)

from nanocobs.

An example:

char buf[16384];
unsigned encoded_len;
fill_buf_width_cobs_encoded_data(buf, &encoded_len); // your function
unsigned decoded_len;
cobs_decode(buf, encoded_len, buf, sizeof(buf), &decoded_len);
// at this point,  "buf" will contain the decoded payload, and "decoded_len" is the length of the decoded payload.

Is this what you're looking to achieve?

from nanocobs.

Yes, thank you! Was indeed a bit confused that the cobs_decode also decodes inplace (but to your defense, this is mentioned in the README, I just overlooked it)

from nanocobs.

Glad it does what you want! I'll update the docs and try to think of better names for the _inplace versions; it's clearly a soft spot in the API.

from nanocobs.