- From: Bob Wyman <bob@wyman.us>
- Date: Fri, 24 Mar 2023 13:48:41 -0400
- To: public-swicg@w3.org
- Message-ID: <CAA1s49XvME8hmH4w8egi5iThGRZWVTHrxzTH+fvHoB9uaudqFQ@mail.gmail.com>
Elsewhere, we started talking about issues with Content Negotiation as
required by ActivityPub 3.2
<https://www.w3.org/TR/activitypub/#retrieving-objects>. I'm copying the
previous messages here to ensure better retrievability in the future and to
prevent distraction from the main subject of the other thread. If further
comments are made on this subject, please reply to this message, not the
other one.
---------- Forwarded message ---------
From: Bob Wyman <bob@wyman.us>
Date: Fri, Mar 24, 2023 at 12:42 PM
Subject: Re: Should the specs be forked and maintained elsewhere?
Melvin, Ryan,
Could you provide at least a short summary of why "content negotiation" has
not lived up to expectations? It is good to know that it is non-optimal. It
would be better to know why.
bob wyman
---------- Forwarded message ---------
From: Melvin Carvalho <melvincarvalho@gmail.com>
Date: Fri, Mar 24, 2023 at 1:14 PM
pá 24. 3. 2023 v 17:42 odesílatel Bob Wyman <bob@wyman.us> napsal:
> Melvin, Ryan,
> Could you provide at least a short summary of why "content negotiation"
> has not lived up to expectations? It is good to know that it is
> non-optimal. It would be better to know why.
>
Hard to configure, hard to debug, hard to make backwards compatible with
the existing web.
It's not easy to cache globally because each document have different
etags. It can open some security holes for servers not configured properly.
Below is a sample of what Dan Brickley (original chair of the social web
XG, author of FOAF, runs schema.org) has to say on content negotiation.
You could also repeat the search for conneg.
e.g. "the content negotiation aspect of linked data has been massively
oversold"
https://twitter.com/search?q=%40danbri%20content%20negotiation&src=typed_query
Id rather it was optional that required.
---------- Forwarded message ---------
From: Ryan Barrett <public@ryanb.org>
Date: Fri, Mar 24, 2023 at 1:16 PM
Sure! And feel free to redirect to a different venue if we're getting off
topic. Off the top of my head:
1. Most web developers are not aware that conneg exists. This is usually
fine, since it's uncommon, but can be an unpleasant surprise when they
first hit it and don't understand it.
2. The `Accept` header is notoriously complicated to parse, generate,
and fully comply with. Imo much of that complexity is unneeded in practice.
3. It often breaks caching when it's first introduced. Internal and
external caches generally ignore Content-Type and Accept (etc) by default,
so when conneg is first introduced, clients often get the wrong response
type. Developers learn the hard way that they need to add Accept to Vary
and to their framework's cache.
4. URLs leak across contexts. They may start out within a single service
or API, but they gradually proliferate into user-visible links, object
identifiers in other systems, add-on services, third party tools, etc.
These generally don't know the original conneg requirements, which causes
unpleasant surprises.
I think most of this boils down to: modality generally considered harmful.
When something always behaves the same way, it's reliable and easy to use.
When it behaves differently based on something far away in the environment
that you may not know exists, it's unreliable and surprising. Add in a very
large ecosystem of independent tools that all need to interoperate, often
in fine-grained ways, and you have a recipe for failure.
---------- Forwarded message ---------
From: Aaron Gray <aaronngray@gmail.com>
Date: Fri, Mar 24, 2023 at 1:22 PM
On Fri, 24 Mar 2023 at 16:49, Aaron Gray <aaronngray@gmail.com> wrote:
> The standards are there for a reason we should be following them.
> Providing access to optional extra typed fields on a post is not complex.
>
To clarify I was referring to JSON-LD and @Content, not conneg HTTP
Content-Type, Accept content negotiation.
Regards,
Aaron
---------- Forwarded message ---------
From: Benjamin Goering <ben@bengo.co>
Date: Fri, Mar 24, 2023 at 1:23 PM
http content negotiation is normatively required by the activitypub spec
(3.2).
I'll also point out that some URIs from the TRs already support it.
```
> bengo@bengo ~ ⚡ curl -H "Accept: application/json"
> https://www.w3.org/ns/activitystreams | head -n 10
> % Total % Received % Xferd Average Speed Time Time Time
> Current
> Dload Upload Total Spent Left
> Speed
> 100 7985 100 7985 0 0 39142 0 --:--:-- --:--:-- --:--:--
> 40125
> {
> "@context": {
> "@vocab": "_:",
> "xsd": "http://www.w3.org/2001/XMLSchema#",
> "as": "https://www.w3.org/ns/activitystreams#",
> "ldp": "http://www.w3.org/ns/ldp#",
> "vcard": "http://www.w3.org/2006/vcard/ns#",
> "id": "@id",
> "type": "@type",
> "Accept": "as:Accept",
> bengo@bengo ~ ⚡ curl -H "Accept: text/html"
> https://www.w3.org/ns/activitystreams | head -n 10
> % Total % Received % Xferd Average Speed Time Time Time
> Current
> Dload Upload Total Spent Left
> Speed
> 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:--
> 0<!DOCTYPE html><html dir="ltr" typeof="bibo:Document " prefix="bibo:
> http://purl.org/ontology/bibo/ w3p: http://www.w3.org/2001/02pd/rec54#
> as: https://www.w3.org/ns/activitystreams#" lang="en"><head><meta
> charset="utf-8"><meta name="generator" content="ReSpec 20.7.1"><meta
> name="viewport" content="width=device-width, initial-scale=1,
> shrink-to-fit=no"><meta property="dc:language" content="en" lang="">
> <title>ActivityStreams 2.0 Terms</title>
>
>
>
> <style>/*****************************************************************
> * ReSpec 3 CSS
> * Robin Berjon - http://berjon.com/
> *****************************************************************/
>
> 100 33068 0 33068 0 0 148k 0 --:--:-- --:--:-- --:--:--
> 153k
> curl: (23) Failure writing output to destination
>
```
--------- Forwarded message ---------
From: Melvin Carvalho <melvincarvalho@gmail.com>
Date: Fri, Mar 24, 2023 at 1:35 PM
Subject: Re: Should the specs be forked and maintained elsewhere?
pá 24. 3. 2023 v 18:23 odesílatel Benjamin Goering <ben@bengo.co> napsal:
> http content negotiation is normatively required by the activitypub spec
> (3.2).
> I'll also point out that some URIs from the TRs already support it.
>
How do you ensure that each accept header gives back the same triple?
For example, do they in this case.
Not a pedantic point, I've been trying to fix this for years in Solid, and
its made harder by the fact that the developers dont know how it works. In
fact the maintainer of the reference implementation said to me recently:
"What's content negotiation". It's hard to fix these things.
---------- Forwarded message ---------
From: Melvin Carvalho <melvincarvalho@gmail.com>
Date: Fri, Mar 24, 2023 at 1:41 PM
pá 24. 3. 2023 v 18:34 odesílatel Melvin Carvalho <melvincarvalho@gmail.com>
napsal:
>
>
> pá 24. 3. 2023 v 18:23 odesílatel Benjamin Goering <ben@bengo.co> napsal:
>
>> http content negotiation is normatively required by the activitypub spec
>> (3.2).
>> I'll also point out that some URIs from the TRs already support it.
>>
>
> How do you ensure that each accept header gives back the same triple?
>
> For example, do they in this case.
>
> Not a pedantic point, I've been trying to fix this for years in Solid, and
> its made harder by the fact that the developers dont know how it works. In
> fact the maintainer of the reference implementation said to me recently:
> "What's content negotiation". It's hard to fix these things.
>
Let me add an extra point
Pretty much all the conneg ive seen is broken (e.g. on mastodon too)
It's generally broken on vocabs ive seen too. But nobody actually cares.
Because, people dont actually use vocabs. They may in future e.g. for
xsd:typing but they are barely noticeable today. The whole overhead of the
thing I would argue provides very little developer value other than a
documentation where they can copy and paste.
Received on Friday, 24 March 2023 17:49:08 UTC