[Feeds] Re: Should we standardize on a single URL path to RSS feeds?

["Martin J. Dürst" <[e-mail]>]

Hello Mark,

On 2024-12-26 22:13, Mark wrote:
> Thanks for the clarification, Martin.
> 
> Hmm, "robots.txt" seems fine, despite the bad little bots that ignore it ;)

It seems fine for a single one in isolation. But if there are many more 
(and there are, see 
https://www.iana.org/assignments/well-known-uris/well-known-uris.xhtml#well-known-uris-1) 
it gets more complicated. When you (plan to) do something, it's often 
good to think about what would happen if everybody (or at least a lot of 
people) did it.

> Maybe I'm missing it. But I'd love to understand why IETF has concluded that fixed URLs are a bad idea, if possible. If anyone can link me to the rationale, that would be great!

The rationale is pretty much given in the Introduction of RFC 8615, in 
particular in paragraph 4. But let me expand a bit more.

Imagine a webmaster looking at the directory of a website. How do they 
decide what files are part of the website, and what files are there 
because of some other reason (such as robots.txt)? How do they come up 
with filenames that make sense for the website, but don't step on 
special-purpose designated filenames? What if the website contains 
information about robots (naturally served in a file robots.html) and 
wants to provide their information also in plain text form (which would 
be named robots.txt, if not for...).

And many websites are not managed by webmasters looking at files in 
directories, but with web frameworks (think Ruby on Rails,...), content 
management systems, and so on. Features that these systems introduce, 
conventions for naming things on the website, and the names of 
special-purpose files such as robots.txt would collide at unpredictable 
times with unpredictable and difficult to debug consequences.

Putting (almost) all the special files into .well-known (I used 
".wellknown" in my last mail, but there's a hyphen) solves the problem 
quite elegantly. Webmasters, frameworks and CMSes have to treat only one 
directory as special.

Hope this helps.

Regards,    Martin.


> Mark Kennedy
> Engineer
> openrss.org
> 
> 
> On Thursday, December 26th, 2024 at 3:04 AM, Martin J. Dürst <[e-mail]> wrote:
> 
>> Hello Mark,
>>
>> There are people on this list with more background knowledge in this
>> area, but robots.txt was probably the earliest example of a fixed link.
>> Then came favicon.ico. These are historic artefacts that are difficult
>> to remove.
>>
>> After favicon.ico, people in the IETF, the W3C, and elsewhere started to
>> realize that this was a bad idea and created explanatory texts such as
>> RFC 3205 (later updated to RFC 9205), and mechanisms such as .wellknown
>> (RFC 5785, later updated to RFC 8615).
>>
>> .wellkown essentially makes the mistake of defining a fixed link
>> (prefix) for one last time, and then stuffs all future fixed links into
>> the /.wellknown/ directory, governed by a registry at IANA to avoid
>> collisions.
>>
>> Regards, Martin.
>>
>> On 2024-12-26 16:26, Mark wrote:
>>
>>> Hey Martin,
>>>
>>> Thanks for the link. I've taken a look at that and it says
>>>
>>> "specifications for applications that use HTTP won't contain fixed application URLs or paths; while it is common practice [...], doing so in an IETF specification is inappropriate".
>>>
>>> This suggests that my original question around a standard and fixed URL path for RSS feeds would be inappropriate for the IETF. However, it contradicts something like section 2.3 of RFC 9309 (https://www.rfc-editor.org/rfc/rfc9309.html#name-access-method) where a fixed URL path is clearly being used for a "robots.txt" file. Please correct me if I'm misunderstanding, but why is having a fixed URL path for something like "robots.txt" acceptable for the IETF, but not for a fixed URL path for RSS feeds?
>>>
>>> Thanks for any clarification.
>>>
>>> Mark Kennedy
>>> Engineer
>>> openrss.org
>>>
>>> On Thursday, December 26th, 2024 at 1:36 AM, Martin J. Dürst [e-mail] wrote:
>>>
>>>> Sorry for being so late, but when cleaning up my inbox, I found this and
>>>> thought that for some people, a look at RFC 9205 (BCP 56), in particular
>>>> Section 4.4
>>>> (https://www.rfc-editor.org/rfc/rfc9205#name-specifying-urls) might
>>>> provide some additional background.
>>>>
>>>> Regards, Martin.
>>>>
>>>> On 2024-12-18 06:41, Rob Sayre wrote:
>>>>
>>>>> Oh, it's so much worse than you think. :)
>>>>>
>>>>> Check this one out:
>>>>>
>>>>> https://help.apple.com/itc/podcasts_connect/#/itcb54353390
>>>>>
>>>>> Look at "itunes:new-feed-url" there. Apologies in advance, it's like
>>>>> nails on a chalkboard.
>>>>>
>>>>> thanks,
>>>>> Rob
>>>>
>>>> --
>>>> Feeds mailing list -- [e-mail]
>>>> To unsubscribe send an email to [e-mail]

-- 
Prof. Dr.sc. Martin J. Dürst
Department of Intelligent Information Technology
College of Science and Engineering
Aoyama Gakuin University
Fuchinobe 5-1-10, Chuo-ku, Sagamihara
252-5258 Japan