3 ms·
I get your point that yes, you could use ASN.1 in a way that isn't extensible. To pick a nit in your nit though: OIDs are arranged in a hierarchy, so anybody w
by tialaramex 7y ago
I get your point that yes, you could use ASN.1 in a way that isn't extensible. To pick a nit in your nit though:
OIDs are arranged in a hierarchy, so anybody who already has an arc can just delegate you an arc from their part of the hierarchy, much the same way foo.example could delegate kragen.foo.example to you, and then you could delegate tialaramex.kragen.foo.example from that to me and so on.
The IETF hijacked the 1.3.6.1 arc, by basically arguing that if hypothetically the US Department of Defence were asked to delegate them the 1.3.6.1 arc from its 1.3.6 arc it would doubtless agree, and so no need to actually ask...
For example 1.3.6.1.5.5.7.3.1 is "serverAuth" the OID used to indicate that an X.509 certificate is intended to identify a TLS Server, that was assigned by the IETF's PKIX many years ago.
But the IETF has an arc under 1.3.6.1 for giving out OID arcs to anybody who needs one, 1.3.6.1.4.1, you can just fill out a form for IANA explaining why you need one (e.g. to make some new SNMP features work) and you'll get an arc back typically within a few days. e.g. 1.3.6.1.4.1.44947 is ISRG, the charity which provides the Let's Encrypt service.
Colliding HTML tag names has been a problem, albeit a small one, even inside "standard" HTML itself because of the conflict between WHATWG and W3C for years. The small number of web browsers with significant market share made this less serious than it might be. As to people redefining MIB OIDs, I can't help idiots, I would regard every instance of this happening as a bug in the later definition.
- kragen 7y agoWhat I mean about ASN.1 is that if you have a field foo that's defined as foo SEQUENCE { corge INTEGER, grault OBJECT IDENTIFIER } you can't compatibly augment it with a new field quux as foo SEQUENCE { corge INTEGER, grault OBJECT IDENTIFIER, quux OBJECT IDENTIFIER } and it doesn't help you at all that grault and quux are both OBJECT IDENTIFIERs. Which HTML tag names had collisions? I can't think of any at the moment, but admittedly I'm no HTML expert, though I've been reading HTML specs since HTML+ and before. I'm more interested in problems not happening than on having someone to blame the problems on.
- tialaramex 7y ago<cite> is the canonical HTML tag collision. WHATWG and W3C defined this differently. In practice as so often in the Web it's resolved by a mixture of just accepting everything and so nobody gets what they wanted, and by semantics of the document being ignored anyway. Yes, you're correct that you can't just shove arbitrary things into ASN.1 in this way and so ASN.1 itself is arguably not where the extensibility is, that's fair.
- kragen 7y agoThanks for the note on <cite>! I don't think that's actually a collision; <cite> was in HTML 2, so there wasn't much time for people to start using it in conflicting ways. What happened with <cite> was that, for about four years, Hixie published a changed definition in the HTML5 spec, one which didn't allow citing people's names, just the titles of their works. This attempted change was eventually reverted. The URL-based and OID-based extensibility mechanisms we're discussing don't even attempt to prevent such problems. What I mean by "collision" is something like the following: one person (say, Hixie) defines <c> in their browser or stylesheet to mean "canonical name" while another person (say, the W3C) defines <c> to mean "category", resulting in problems when you try to view documents written for one browser using the other browser. I don't know of anything like this having happened in HTML, ever.
- zzo38computer 7y agoFortunately there is a "2.25." OID block, for use to assign a OID from a UUID, without having to call anyone else. But is there a similar thing that can be done to make a PostScript XUID from a UUID? PostScript supports only 32-bit integers and not 128-bit integers, but you can use four 32-bit integers, with some prefix.