On 22. 02. 21 10:49, Martin Björklund wrote: > Hi, > > Section 11 of RFC 7950 says: > > o A "type" statement may be replaced with another "type" statement > that does not change the syntax or semantics of the type. For > example, an inline type definition may be replaced with a typedef, > but an int8 type cannot be replaced by an int16, since the syntax > would change. > > If we're just considering XML, then the syntax or encoding wouldn't > change if we went from > > type int64 { range "2..4"; } > > to > > type string { pattern "2|3|4"; } > > or > > type enumeration { > enum 2; > enum 3; > enum 4; > } > > or > > type union { > type uint8 { range "2"; } > type string { pattern "3"; } > type enumeration { enum 4; } > } > > > But I don't think this is reasonable, and not the intention. I think > that changing the base built-in type always should be considered > non-backwards compatible (which the quoted text above seems to imply).
Agreed. Another problem related to sec. 11 is that it permits to update a module so that the range specification is extended, which may then expose the incompatibility of e.g. uint8 and int8. But I thought that Jürgen's question was directed to the definition of backward compatibility in the semver context. Lada > > > /martin > > > > > Juergen Schoenwaelder <j.schoenwael...@jacobs-university.de> wrote: >> On Fri, Feb 19, 2021 at 10:32:34PM +0100, Carsten Bormann wrote: >>> >>> >>>> On 2021-02-19, at 19:18, Juergen Schoenwaelder >>>> <j.schoenwael...@jacobs-university.de> wrote: >>>> >>>> I think the CBOR encoding picks different tags depending on the >>>> signedness of the base type and this is why things are not that simple >>>> anymore. >>> >>> (This is not the CBOR encoding, but the COMI encoding of keys in URIs.) >> >> OK. The CBOR document indeed says: >> >> 6.1. The unsigned integer Types >> >> Leafs of type uint8, uint16, uint32 and uint64 MUST be encoded using >> a CBOR unsigned integer data item (major type 0). >> >> 6.2. The integer Types >> >> Leafs of type int8, int16, int32 and int64 MUST be encoded using >> either CBOR unsigned integer (major type 0) or CBOR negative integer >> (major type 1), depending on the actual value. >> >> This means the type 'int8 { range 0..10; }' leads to the same >> encodings as the type 'uint8 { range 0..10; }'. >> >>>> For the XML and JSON encodings, all definitions lead to the >>>> same on-the-wire representation, hence the difference is more an >>>> implementation detail. I have no clue what the gnmi people do. The >>>> more diverse encodings we add, the more complex things get. >>> >>> Well, if the equivalence expectation that I was trying to describe actually >>> is ingrained, then whoever designs an encoding (COMI for its URI encoding >>> included) needs to respect it. That would be important to know. >>> >> >> Exactly. I think we never defined this. And of course, this can get >> even more fun if you consider string based encodings. The type >> >> type string { pattern "1|2|3|4"; } >> >> yields the same _XML encoded_ value space as >> >> type int32 { range "1..4"; } >> >> but as far as I recall the JSON/CBOR encodings will treat these two >> differently. So yes, ideally the YANG language would have clear rules >> what YANG's type equivalences are. >> >> /js >> >> -- >> Juergen Schoenwaelder Jacobs University Bremen gGmbH >> Phone: +49 421 200 3587 Campus Ring 1 | 28759 Bremen | Germany >> Fax: +49 421 200 3103 <https://www.jacobs-university.de/> >> >> _______________________________________________ >> netmod mailing list >> netmod@ietf.org >> https://www.ietf.org/mailman/listinfo/netmod > > _______________________________________________ > netmod mailing list > netmod@ietf.org > https://www.ietf.org/mailman/listinfo/netmod > -- Ladislav Lhotka Head, CZ.NIC Labs PGP Key ID: 0xB8F92B08A9F76C67 _______________________________________________ netmod mailing list netmod@ietf.org https://www.ietf.org/mailman/listinfo/netmod