Encoding Repeated Elements Discussion Wrong

[kevin-314159]

I'm brand new to Protobuf, so maybe I have this all wrong, but as best as I could make out, including by experimentation using protoc, the documentation on the encoding of repeated elements is wrong.

It says that if we "construct a Test4 message with d set to "hello", and e set to 1, 2, and 3, this could be encoded as 3206038e029ea705, or written out as Protoscope:"

4: {"hello"}
6: {3 270 86942}

First, the encoding given appears to be just for the values of e, not for the entire Test4 message.
Second, it appears the encoding is for e being set to 3, 270, and 86942, not 1, 2, and 3.

[turkycat]

I'm also reading the documentation for the first time and discovered the same issue. The author may have intended to use varints of length 1, 2, and 3 (because that is the byte length of the numbers 3, 270, and 86942 encoded as varints).

The encoding checks out, but the language isn't clear.

I've created #294 to resolve the issue.

encoding walkthrough

3206038e029ea705

0x32 = 0b 0011 0010
or 0110 (field number 6) 010 (wire type LEN)

06 = length of the payload

Value 1: 03
0000 0011 (MSB=0, done)
= 3

Value 2: 8e 02
1000 1110 (MSB=1, continue) = 000 1110 = 14
0000 0010 (MSB=0, done) = 000 0010 = 2

Value = 14 + (2 << 7) = 14 + 256 = 270

Value 3: 9e a7 05
1001 1110 (MSB=1, continue) = 001 1110 = 30
1010 0111 (MSB=1, continue) = 010 0111 = 39
0000 0101 (MSB=0, done) = 000 0101 = 5

Value = 30 + (39 << 7) + (5 << 14)
= 30 + 4992 + 81920
= 86942