Do we also want to allow single quotes (') here?

I'm not sure single quotes are allowed, but if others agree I can add it to this PR.

No, single quotes should not be interpreted as string delimiters, see definition of quoted-string in RFC 723 §3.2.6. Most other implementations I've looked at don't interpret single quotes. For example: Go test line 196

Don't a lot of HTTP headers have the standard value; optional-key="value" format? Is there an RFC or standard of these headers - or even an unofficial standard - to implement a parser for them?

@RX14 There are a ton of RFCs about mime type syntax. The most recent one is RFC 7321.

As already mentioned, I have a working parser implemented, it just needs #5765 and a little bit of finishing 👍

I'm not talking about mime types - i'm talking about the generic HTTP header value syntax. We already need this for here and thats nothing to do with mime types. The syntax dates from the email header days.

Yeah, I know what this is about. It's not specific to any protocol, neither HTTP nor mail. I'm not sure if there's a proper name, but this is commonly described as mime type (examples: .NET, Java, Rust, Go)

And despite any of this, it's still true that Crystal needs a proper parser implementation.

My bad, that was mime types; i'm a bit tired. I meant that mime types, set-cookie, and content-disposition have the same approximate header syntax with one or more initial values then the same primaryvalue; optionalkey=value; optionalkey=value format. Does that syntax have a name and should we have a generic parser for HTTP headers of this style?

As far as I know, this is called mime type, despite the fact that applications like Set-Cookie have nothing to to with mime. And yes, we should have a generic parser for this.

I've never seen it named as such and I can't imagine it being called that since it would be so confusing. "mime type" already means a specific concrete RFC and concept.

Sry, I meant media type. Doesn't make any difference, though. I guess it's just that this concept initially comes from mime type declarations and was adopted for other uses as well, without there being a proper name for this syntax.

Implementation wise, it is typically named a MIME media type parser. This implementation can be applied to non-MIME contexts such as Set-Cookie or they can use a more specific, custom parser. Go for example uses ParseMimeType for Content-Type, Content-Disposition but a custom implementation which only recognizes a fixed set of attributes for Set-Cookie.

I'd never find such a parser as a user if I had to think "mime". So I think we shouldn't name it that way. It should be in the generic HTTP module.

Merging the fix. Discussion feels unrelated or that it could come later.

Yep, discussion was unrelated, sorry.