RR type encoding length

[pusateri]

Currently, the RR type encoding length is set to the length of the RR type data plus 2. RR type A has length 6, RR Type AAAA has length 18, etc. I think this includes the RDATA LEN short but it seems like it shouldn't.

[pusateri]

If you agree this should be changed, I can submit a patch that changes all of the RR types to return the RDLEN and write the RDLEN in question.encode() and answer.encode() instead of in each RR type encoder.

[silverwind]

Yes, this seems a bit odd. Do we actually need these encodingLength properties? Why not just buf.length?

[pusateri]

encode() uses them to calculate the buffer length to allocate:
L908
Which means it runs the same code twice for name and string encoding to calculate the length the first time and to copy into the buffer the second time. This seems like a lot of work to pre-calculate the length of the Buffer. Passing in a buffer to encode() gets around this issue and then none of the encodingLength code is used.

[pusateri]

But this is secondary to the problem that the encoders for each RR type include the extra RDLEN bytes.