Is this codec in unsigned-varint format?

[keepsimple1]

I must have missed something obvious. For example, this item in the current table:

leofcoin-block,                 ipld,           0x81,           draft,     Leofcoin Block

The code 0x81 is binary 10000001, and in my understanding of unsigned varint, should indicate there is a 2nd byte. Why isn't there 2nd byte in the table for this codec?

[vmx]

The table lists the un-encoded numbers. So in this case 0x81 would mean that if it is e.g. used in a CID, the unsigned varint of it would be 0x8101.

[keepsimple1]

May I suggest adding (or just replacing with) the actual unsigned varint encoded code? Because in the real life uses, we are using and seeing encoded numbers.

Just like UTF-8 which uses similar technique, we only care about the actual encoded bytes.

[rvagg]

I'm not sure I can think of a case where I've had to manually type out the encoded form, much more typically I'm using the integer form, e.g. https://github.com/ipld/js-dag-json/blob/328eda133620a0aff8cb19a314b5622f0054a6d3/index.js#L198 and I'm regularly copy and pasting from this table so it'd be a massive loss to ditch the human-readable forms for computer-redable ones.

You might be able to make a case for tacking an additional column with the bytes, but we're not dealing with a very extensible format with our CSV and haven't made any progress on a more extensible form unfortunately.

[fabianhjr]

Hey, @Stebalien I am currently working on some implementations and would appreciate some clarification about this issue.

For the example @keepsimple1 raised (0x81) would it be to take that value (129 in decimal) and then encode it as an Unsigned Var Int (hence the 0x8101 given by @vmx)?

Are the entries then in big endian notation? (So udp multiaddr which is noted as 0x0111 would then be 273 and would then be 0x9102 as an Unsigned Var Int?

[rvagg]

1. Yes, if encoding 0x81 from the table as a varint, then you'd get two bytes, 0x81 and 0x01, or 129 & 1. The multicodec table doesn't dictate the use of varints which is why it's listed as an integer there. Varints are a use-case implementation detail - CIDs (only one use-case of the multicodec table) do use varints and we generally expect they're used because they're compact and let us span a large range without worrying too much about lower, more common values taking up a lot of space.
2. Have a look at https://github.com/multiformats/unsigned-varint which attempts to describe what "varint" means in the multiformats context, with links to compatible implementations in JS, Rust and Go that you can compare against.