Re: [Rfc-markdown] How to cite this DOI?

[Carsten Bormann <[e-mail]>]

On 15. Oct 2021, at 23:02, John Levine <[e-mail]> wrote:
> 
> It appears that Carsten Bormann  <[e-mail]> said:
>> On 15. Oct 2021, at 12:36, Lars Eggert <[e-mail]> wrote:
>>> 
>>> I have no idea. The whole DOI thing seems pretty sloppily put together; unfortunately it's what the publishing industry adopted though.
>> 
>> That was easy:
>> 
>> https://www.doi.org/doi_handbook/2_Numbering.html#2.2.1
>> 
>>>> The DOI name is case-insensitive and can incorporate any printable characters from the legal graphic characters of Unicode. Further constraints on
>> character use (e.g. use of language-specific alphanumeric characters) can be defined for an application by the ISO 26324 Registration Authority.
>> 
>> You can’t make this stuff up.
> 
> How soon they forget.  DOIs are an instance of the handle system, whose syntax is defined in RFC 3651.  It does indeed allow any UTF-8 encoded
> Unicode 2.0 character, but since the handle system also had its own retrieval protocol, quoting wasn't a problem.

Does “printable character” include ZWNJ?  If not, how do I get a correct representation of the Auf‌lage (*) of a digital object?
The term “case-insensitive” doesn’t even mean anything without a locale.
(Well, you know all this, but I want to remind everyone why this choice just boggles the mind.)

> Except that nobody ever used the handle protocol so the only popular application of handles, DOI, made an expedient
> choice to use http instead, turning the handles into URLs.  Unfortunately, the DOI people are librarians, not
> computer standards experts and may have missed the fine points of what goes into a URL.

The URL representation is not the problem here.

> Having said that, what's the problem with that URL?  RFC 1738 specifically says that unencoded parentheses in a URL are allowed.
> I tried pasting it into Brave (based on Chrome), Safari, and Firefox, and it worked fine.  If you have some tool that barfs
> on them, you can always percent encode them.

Parentheses are just fine in a URL (e.g., I use them all the time on cbor.me [1]).
Everything is wrong about using anything but [-A-Za-z0-9.]+ in the components of a DOI [2]; Elsevier would not have had to do that.

                    .oOo.

So the payload of this message is that I have added the parentheses for the YAML prelude in 1.5.10, but we have to be a bit careful about what is allowed directly in {{}} (right now, once you go outside sequences of XML namechars, weird things will happen).  Because the RFC editor surely wouldn't accept anything else as a reference anchor, I don’t think this is a crippling limitation.

If you ever need more bugs 𐲃𐲡 in your DOI, just holler.

Grüße, Carsten

(*) literally print run, but often used in place of edition, look closely at how this does not have an fl ligature even in a font that has them.
[1]: http://cbor.me/?bytes=82(01-82(02-03))
[2]: https://en.wikipedia.org/wiki/Principle_of_least_astonishment