Validate cookie name prefixes

[Blacksmoke16]

Validates cookie names with prefixes as per https://datatracker.ietf.org/doc/html/draft-ietf-httpbis-rfc6265bis-05#section-4.1.3.

I'm also not really sure why the other name/value validation raises an IO::Error? Seems more appropriate as an ArgumentError.

[Sija]

Pedantic: whitespaces

[straight-shoota]

I'm doubting this is a good change in terms of API design. Now validate_name not just validates the name itself but in the context of other values. This creates some implicit restrictions when using the API in a builder flow, enforcing a specific order of property assignments.

I would suggest to add a dedicated method to validate the entire cookie. This validation could be applied automatically in parse_set_cookie.

[Blacksmoke16]

I would suggest to add a dedicated method to validate the entire cookie. This validation could be applied automatically in parse_set_cookie.

@straight-shoota parse_set_cookie wouldn't handle all the ways you could create a cookie, e.g. when creating it via .new or using the name setter.

This creates some implicit restrictions when using the API in a builder flow, enforcing a specific order of property assignments.

I don't see how this could be avoided. Otherwise you could "build" an invalid cookie and wouldn't even know it.

[straight-shoota]

The answer is that in order to validate the cookie value, you need to validate explicitly when building an instance. I don't think a setter should apply implicit validation on an incomplete cookie.

[Blacksmoke16]

@straight-shoota Oh I see. You were proposing this method be public, then just also call it in the constructor. What are your thoughts on how it should work then? Like .valid? : Bool, or .validate! : Nil | NoReturn, or?

[straight-shoota]

Prob. both?

[sdogruyol]

Thank you @Blacksmoke16 🙏