Implement is_human_readable to return false

[CBenoit]

Currently, is_human_readable is not manually implemented and thus returns true (default). As a consequence, types supporting a more compact binary representation are serialized inefficiently (typically into a string representation).

This is a breaking change: previously serialized types supporting a compact representation will fail at deserialization with this commit.

This is an issue preventing adoption by some projects.

[KizzyCode]

I see. As far as I understand, this is only an issue for formats that use multiple intermediate representations during serialization: T → serde_asn1_der supported native type → DER?

E.g. a date type which could either use:

• Date <=> Human-Readable String <=> DER UTF-8 String
• Date <=> Compact String <=> DER UTF-8 String

The intermediate representation is selected by is_human_readable, and if we change this, another intermediate representation and subsequently a different (de-)serialization chain is selected which could break compatibility?

[CBenoit]

I see. As far as I understand, this is only an issue for formats that use multiple intermediate representations during serialization: T → serde_asn1_der supported native type → DER?

E.g. a date type which could either use:

* `Date <=> Human-Readable String <=> DER UTF-8 String`

* `Date <=> Compact String <=> DER UTF-8 String`

The intermediate representation is selected by is_human_readable, and if we change this, another intermediate representation and subsequently a different (de-)serialization chain is selected which could break compatibility?

Correct.

It depends on the Serialize and Deserialize implementation of the type itself. See for instance this Serialize implementation for an IPv4 address type.
An IPv4 address takes 4 bytes to encode in non human readable binary format, but between 7 and 15 bytes when encoded as characters (x.x.x.x - xxx.xxx.xxx.xxx).

This might or might not cause compatibility issues depending on whether downstream user used such a type. The library version number should be bumped appropriately.

[KizzyCode]

Merged and published. Thank you very much! 😊