(Question and For the record) Why is Client Identifier Prefix using colon ":" as a delimiter?

[sakimura]

I guess it is because it started as a scheme but I was not sure.
The use of ":" as the delimiter for Client Identifier Prefix removes the possibility of using urn as the prefix.
Is the choice of ":" as the delimiter intentional?

[jogu]

It was a somewhat arbitrary choice from what I remember, though it at least was one that's not already commonly used in generated client ids. Is there a character that would work better?

[bc-pi]

I mentioned some distaste for the use of ":" as the delimiter in #376 (comment)

"|" or the url safe "~" or maybe some combination of two or there url safe characters ".." or "~~" or "~~" or ".~." etc. might be "better" but my sense is that there's not the appetite for a breaking change there at this point.

[sakimura]

Given that clients not updated may send URI-based client_id such as did: or https: etc., I thought a character which is not allowed in the URI scheme part might be better. To allow UUID and URN as a prefix, "-" and ":" should be avoided as well. "|" or "~" or even "_" would probably work.

[sakimura]

might be "better" but my sense is that there's not the appetite for a breaking change there at this point.

I guess NOW is the last chance to fix it. I support your proposal.

[bc-pi]

might be "better" but my sense is that there's not the appetite for a breaking change there at this point.

I guess NOW is the last chance to fix it. I support your proposal.

Perhaps you can convince the WG and chairs otherwise but, the way I see it, the chance to fix it is long past; when ":" went in as the prefix delimiter. Even back then I did mention that a different delimiter could, perhaps should, be used #263 (review) but that piece seems to have been lost in the discussion about which things to promote or prescribe as preferred mechanisms by giving them special treatment.

An immediate actionable way you could support this is to remove the request for changes you have on #401

[sakimura]

Aw. I came in too late 😭 2025年4月7日(月) 2:06 Brian Campbell ***@***.***>:

…

might be "better" but my sense is that there's not the appetite for a breaking change there at this point. I guess NOW is the last chance to fix it. I support your proposal. Perhaps you can convince the WG and chairs otherwise but, the way I see it, the chance to fix it is long past; when ":" went in as the prefix delimiter. Even back then I did mention that a different delimiter could, perhaps should, be used #263 (review) <#263 (review)> but that piece seems to have been lost in the discussion about which things to promote or prescribe as preferred mechanisms by giving them special treatment. An immediate actionable way you could support this is to remove the request for changes you have on #401 <#401> — Reply to this email directly, view it on GitHub <#505 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABFENZSJHTZLFBBPYGQLKL2YFNJJAVCNFSM6AAAAAB2O5LGFOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDOOBRGUYTINZZGI> . You are receiving this because you authored the thread.Message ID: ***@***.***> [image: bc-pi]*bc-pi* left a comment (openid/OpenID4VP#505) <#505 (comment)> might be "better" but my sense is that there's not the appetite for a breaking change there at this point. I guess NOW is the last chance to fix it. I support your proposal. Perhaps you can convince the WG and chairs otherwise but, the way I see it, the chance to fix it is long past; when ":" went in as the prefix delimiter. Even back then I did mention that a different delimiter could, perhaps should, be used #263 (review) <#263 (review)> but that piece seems to have been lost in the discussion about which things to promote or prescribe as preferred mechanisms by giving them special treatment. An immediate actionable way you could support this is to remove the request for changes you have on #401 <#401> — Reply to this email directly, view it on GitHub <#505 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABFENZSJHTZLFBBPYGQLKL2YFNJJAVCNFSM6AAAAAB2O5LGFOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDOOBRGUYTINZZGI> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[danielfett]

If we don't want to use :, it should at least be a url-safe character like ~ or !.

[sakimura]

I am fine with ~, !. Actually, "|" can be interpreted by some parser as a pipe so it may be bad in that respect as well.
Also ":" is not URL safe.

[Sakurann]

WG discussion

• it is not worth a breaking change now
• add a note in the spec, clarifying how to process this first colon part.