Re: [Cbor] List of not-well-formed CBOR and test vectors

[Laurence Lundblade <[e-mail]>]

I’ve added:
    - The simple values 24-31, the ones marked as “reserved" in 7409 (I’d prefer to call them “permanently disallowed” or such)
    - Half-precision, float and short values that are missing some bytes

Will turn this in to an appendix for CBORbis in about a week. 

Thanks for any comments, additions, confirmations.

LL



> On Jul 29, 2019, at 11:49 AM, Laurence Lundblade <[e-mail]> wrote:
> 
> I think I’ve made a comprehensive list of all things that are not well formed by going through the latest draft and my decoder. There are about a dozen of them. I’ve also created 110 test vectors that cover them pretty thoroughly. 
> 
> Everything is here <https://github.com/laurencelundblade/QCBOR/blob/not_well_formed/test/not_well_formed_cbor.h> in a C header. The dozen types of non-well-formedness are listed as comments in the header file. The test vectors are in an array that can be used for testing. It is BSD-3 license.
> 
> I’ve turned up one bug in the RFC’s pseudo code. It doesn’t catch an indefinite length string as a segment in another indefinite length string. 
> 
> I’d like to get some review, some folks to try it out and such to see if I’ve missed anything and all is right. When that is done I’ll make a pull request for the draft out it. Probably in about two weeks.
> 
> So please look it over and try it against your implementation.
> 
> LL
>