Re: [Curdle] Time to Review IANA SSH Registries Policies?
Mouse <[email protected]> Sat, 6 Feb 2021 12:28:10 -0500 (EST)
| Newsgroups | gmane.ietf.secsh,gmane.ietf.curdle |
|---|---|
| Message-ID | <[email protected]> |
>> identifiers without going through the registry (specifically, >> identifiers of the form name@domain are permitted, as assigned by >> the owner of that domain). Which, actually, brings up a point. What if the domain in question disappears or changes "owner"s? > It is possible to mark [email protected] as private/experimental and not > require review. Indeed, I think that's exactly what they are: private to the person or organization controlling dom.ain. (But the part before the @ is not necessarily any kind of "user"; see 4250 4.6.1, third paragraph.) > The only issue would be if such identifiers are needed for interop > and the table has to be in IANA. Is that the case? In practice, it is not. There are numerous examples of such names in live use, some of which interoperate (and, except for some that are undocumented, the ones that don't don't because implementors haven't bothered, not because there's anything preventing it), without any sort of IANA interaction required. However, I have seen an ssh server (embedded in a commercial turnkey product - a switch, I think it was) that would crash upon, as far as I could tell, receiving any instance of the [email protected] extension mechanism. I had to add an option to moussh to suppress sending any such in order for it to interoperate with that device. (Not that this is more than a side note; if we don't do things because buggy implementations may mishandle them, we'll never do anything. That was something like a decade ago, but there are recent issues too; I just recently ran into an ssh server that didn't offer either of the kex methods required by 4253 6.5.) Of course, there is nothing technical preventing anyone from implementing something that looks like that extension mechanism _without_ the approval of the holder of the dom.ain in question. But doing that, like any other spec violation, is just _asking_ for interoperability problems. /~\ The ASCII Mouse \ / Ribbon Campaign X Against HTML [email protected] / \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B